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

2: %\linespread{1.6}

3: \shorttitle{The Ca {\footnotesize II} Triplet as Metallicity indicator}

4: \shortauthors{Carrera et al.}

5: 

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

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

8: 

9: \begin{document}

10: \newcommand{\ki}{[Fe/H]$_{KI03}$}

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

12: 

13: 

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

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

16: %% you desire.

17: 

18: \title{The Infrared Ca {\footnotesize II} triplet as metallicity indicator}

19: 

20: 

21: \author{R. Carrera and C. Gallart}

22: \affil{Instituto de Astrof\'{\i}sica de Canarias, Spain}

23: \email{rcarrera@iac.es}

24: \email{carme@iac.es}

25: 

26: \author{E. Pancino}

27: \affil{Osservatorio Astronomico di Bologna, Italy}

28: 

29: \and

30: 

31: \author{R. Zinn}

32: \affil{Department of Astronomy, Yale University, USA}

33: 

34: \begin{abstract}

35: 

36: From observations of almost 500 RGB stars in 29 Galactic open and

37: globular clusters, we have investigated the behaviour of the infrared

38: Ca {\footnotesize II} triplet (8498, 8542 and 8662 \AA) in the age

39: range 13$\leq$Age/Gyr$\leq$0.25 and the metallicity range $-2.2\leq$

40: [Fe/H] $\leq$+0.47. These are the widest ranges of ages and

41: metallicities in which the behaviour of the Ca {\footnotesize II}

42: triplet lines has been investigated in a homogeneous way. We report

43: the first empirical study of the variation of the Ca{\footnotesize II}

44: triplet lines strength, for given metallicities, with respect to

45: luminosity. We find that the sequence defined by each cluster in the

46: Luminosity-$\Sigma$Ca plane is not exactly linear. However, when only

47: stars in a small magnitude interval are observed, the sequences can be

48: considered as linear. We have studied the the Ca {\footnotesize II}

49: triplet lines on three metallicities scales. While a linear

50: correlation between the reduced equivalent width ($W'_V$ or $W'_I$)

51: {\it versus} metallicity is found in the \citet{cg97} and

52: \citet{ki03} scales, a second order term needs to be added when the

53: \citet{zw84} scale is adopted. We investigate the role of age from the

54: wide range of ages covered by our sample. We find that age has a weak

55: influence on the final relationship. Finally, the relationship derived

56: here is used to estimate the metallicities of three poorly studied open

57: clusters: Berkeley 39, Trumpler 5 and Collinder 110. For the latter, the

58: metallicity derived here is the first spectroscopic estimate

59: available.

60: \end{abstract}

61: 

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

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

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

65: %% what keyword punctuation is appropriate.

66: 

67: %% Authors who wish to have the most important objects in their paper

68: %% linked in the electronic edition to a data center may do so in the

69: %% subject header.  Objects should be in the appropriate "individual"

70: %% headers (e.g. quasars: individual, stars: individual, etc.) with the

71: %% additional provision that the total number of headers, including each

72: %% individual object, not exceed six.  The \objectname{} macro, and its

73: %% alias \object{}, is used to mark each object.  The macro takes the object

74: %% name as its primary argument.  This name will appear in the paper

75: %% and serve as the link's anchor in the electronic edition if the name

76: %% is recognized by the data centers.  The macro also takes an optional

77: %% argument in parentheses in cases where the data center identification

78: %% differs from what is to be printed in the paper.

79: 

80: \keywords{stars: abundances --- stars: late-type --- globular clusters: general ---

81: open clusters: individual(\objectname{Berkeley 39},\objectname{Collinder 110},

82: \object{Trumpler 5})}

83: 

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

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

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

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

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

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

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

91: %% each reference.

92: \section{Introduction}

93: 

94: The main functions defining the star formation history of a complex

95: stellar system are the star formation rate, SFR($t$) and the chemical

96: enrichment law, $Z(t)$, both function of time. The SFR($t$) can be

97: derived in detail from deep color--magnitude diagrams. $Z(t)$ has been

98: traditionally constrained by the color distribution of RGB

99: stars. However, this method of deriving metallicities from photometry

100: is a very crude one because in the RGB there is a degeneracy between

101: age and metallicity. To break this degeneracy we may obtain

102: metallicities from another source and then derive the age from the

103: positions of stars in the color--magnitude diagram. Of course, the

104: best way to obtain stellar metallicities is high-resolution

105: spectroscopy, which also provides abundances of key chemical

106: elements. However, a lot of telescope time is necessary to measure a

107: suitable number of stars. The alternative is low-resolution

108: spectroscopy, which allows us to observe a large number of stars in a

109: reasonable time using modern multi-object spectrographs.  At low

110: resolution, the metallicity is obtained from a spectroscopic line

111: strength index. The Mg$_2$, Ca {\footnotesize II} H \& K and Ca {\footnotesize II} infrared

112: triplet lines, the Fe lines, etc., are the most widely used indexes

113: for obtaining stellar metallicities. Different indexes are adequate

114: for different types of stars. For example, Fe lines are useful for

115: stars at the base of the RGB or in the main sequence

116: turn-off. Observation of these stars, however, is only possible for

117: the closest systems and even those require 8 m-class telescopes and

118: long integration times. Thus, for external galaxies, the only stars

119: that can be observed with modern multi-object spectrographs and

120: reasonable amounts of telescope time are those near the tip of the

121: RGB. A good spectroscopic index to obtain metallicities for these

122: stars is the infrared Ca {\footnotesize II} triplet (CaT), whose lines

123: are the strongest features in the infrared spectra of red giant stars.

124: 

125: \citet{az88} demonstrated that in the integrated

126: spectra of Galactic globular clusters, the equivalent widths of CaT

127: lines are strongly correlated with metallicity. As the near-infrared

128: light of globular clusters, where the CaT lines are, is dominated by

129: the red giant contribution, this relation may be also true in these

130: stars individually. Subsequent studies focused on the analysis of

131: individual red giants in globular clusters

132: \citep[e.g.][]{adc91}. These studies demonstrated that the strength of

133: the CaT lines changes systematically with luminosity along the

134: RGB. Moveover, for a given luminosity, the strength of these lines is

135: correlated with the cluster metallicity. Many authors have obtained

136: empirical relationships between the combined equivalent width of the

137: CaT lines and cluster metallicity. A very comprehensive work in this

138: field was published by \citet{r97a}, based on 52 Galactic globular

139: clusters covering a metallicity range of $-2\leq$ [Fe/H]

140: $\leq-0.7$. They compared the resulting calibration in the

141: \citet{zw84} and \citet{cg97} metallicity scales. While in the

142: \citet{cg97} scale a linear correlation between metallicity and

143: equivalent width of the CaT lines at the level of the

144: horizontal-branch (HB) V-V$_{HB}$=0 (known as reduced equivalent

145: width) was found for all clusters, this relationship was not linear

146: when the \citet{zw84} scale was used. In most studies, the run of CaT

147: lines with metallicity has been investigated in globular clusters

148: only, which have all similar ages. If we wish to derive stellar

149: metallicities in systems in which star formation has taken place in

150: the last few Gyr, such as dwarf irregular galaxies or open clusters,

151: it is necessary to address the role of age on the CaT strength. Some

152: authors have used (a few) young open clusters to study the behaviour

153: of the CaT with metallicity \citep[e.g.][]{sunt92}, using the

154: \citet{zw84} metallicity scale as reference. \citet{c04} very recently

155: obtained a new relationship, using open and globular clusters covering

156: $-2\leq$ [Fe/H] $\leq-0.2$ and 2.5 $\leq$ (age/Gyr) $\leq$ 13 in the

157: \citet{cg97} scale. They found a linear correlation among the reduced

158: equivalent width and metallicity. This indicates a weak influence of

159: age in the range of ages investigated (age $\geq$ 2.5 Gyr). However,

160: to apply this relationship to systems with star formation over the

161: last Gyr and/or with stars more metal-rich than the solar metallicity,

162: it is necessary to investigate its behaviour further for younger ages

163: and higher metallicities.

164: 

165: The purpose of this paper is to obtain a new relationship between the

166: equivalent width of the CaT lines and metallicity, covering a range as

167: wide as possible of age and metallicity. Our sample covers $-2.2\leq$

168: [Fe/H] $\leq$+0.47 and 0.25 $\leq$ Age/Gyr $\leq$ 13. The influence of

169: age and the variation of the CaT lines along the RGB are

170: investigated. In Section \ref{sample}, we present the cluster

171: sample. In Section \ref{obsdata}, the observations and data reduction

172: are described. The way in which the equivalent width of the the CaT

173: lines has been computed is described in Section \ref{catriplet}, where

174: the behaviour of the CaT with luminosity is also investigated. In

175: Section \ref{catmetallicityscale} we obtain the relationship between

176: the equivalent width of the CaT lines and metallicity, and we discuss

177: the influence of age and the [Ca/Fe] ratio in them. Finally, the

178: derived relationships are used in Section \ref{derivedmetallicities}

179: to obtain the metallicities of the open clusters Berkeley 39, Trumpler

180: 5 and Collinder 110.

181: 

182: \section{Clusters Sample\label{sample}} 

183: 

184: \placetable{cluster sample}

185: 

186: To study the behaviour of the CaT lines with metallicity, we have

187: observed individual stars, with available V magnitudes, in 29 stellar

188: clusters (15 open and 14 globular). Of the 29 clusters in this sample,

189: 27 also have I magnitudes available. This sample covers the widest

190: range of ages (0.25 $\leq$ Age/Gyr $\leq$1 3) and metallicities (2.2

191: $\leq$ [Fe/H] $\leq$ +0.47) in which the CaT lines have been observed

192: in a homogeneous way. The main parameters of the observed clusters are

193: listed in Table \ref{clustersample}. Our sample covers most of the

194: open clusters visible from the northern hemisphere with enough stars

195: above the red clump to get a good sampling of the RGB, and with

196: magnitudes easily reachable with the INT, WHT and 2.2 m CAHA

197: telescopes. In particular, the sample contains NGC 6705 (M11), a very

198: young open cluster (0.25 Gyr) with a well populated RGB, and NGC 6791,

199: one of the oldest open clusters ($\sim$9 Gyr), which is among the most

200: metal-rich clusters in our Galaxy ([Fe/H] $\sim$ +0.47). From the

201: south, using the VLT\footnote{Based on observations made with ESO

202: telescopes at Paranal observatories under programme 074.B-0446(B).}

203: and CTIO 4 m telescope, we observed four globular clusters, including

204: NGC 5927 and NGC 6528, which are among the most metal-rich globular

205: clusters in our Galaxy. The sample also includes the observations of 9

206: globular and 3 open clusters available at the ESO archive, whose

207: observations were carried out with the same instrumental

208: configurations as our own. With the purpose of investigating the

209: behaviour of the CaT lines with luminosity, we have observed stars

210: along the RGB in 5 clusters spanning our whole range of metallicities.

211: 

212: Table \ref{clustersample} presents a list of all the clusters in our

213: sample, together with their main characteristics: age, distance

214: modulus, reddening, reference metallicities in 3 scales (see Section

215: \label{catmetallicityscale}) and [Ca/H]. In total, 26 of the 29

216: observed clusters have metallicities in at least one of the three

217: scales. For the other 3 clusters (Collinder 110, Trumpler 5 and

218: Berkeley 39), we calculate their metallicities with the relationships

219: obtained here.

220: 

221: \placetable{starsample}

222:     

223: \section{Observations and Data Reduction\label{obsdata}}

224: 

225: About 500 stars have been observed in the 29 clusters of our sample in

226: 6 different runs from 2002 to 2005, using the William Herschel

227: Telescope (WHT) and Isaac Newton Telescope (INT), both at Roque de los

228: Muchachos Observatory (La Palma, Spain), the 4 m telescope at CTIO (La

229: Serena, Chile), the 2.2 m at the Calar Alto Observatory (Almeria,

230: Spain) and the VLT at Paranal Observatory (Chile). The dates,

231: instruments and spectral resolution for each run are listed in Table

232: \ref{runs}. The instrumental configurations have been chosen in order

233: to ensure that the resolution was similar in each run.  The exposure

234: times were selected as a function of the magnitude of the stars in

235: order to obtain a good S/N, which in most cases was greater than

236: 20. We have rejected from the analysis those stars with S/N lower than

237: 20 (see below). In each run we have observed a few stars in common

238: with other runs in order to ensure the homogeneity of our

239: sample. Equivalent widths obtained for each star observed in two or

240: more runs have been plotted in Figure \ref{telescopes}. The

241: differences between runs are $< 0.1 \pm0.1$ \AA. The calculated

242: equivalent widths, together with the obtained radial velocity and the

243: utilized V and I magnitudes, are listed in Table \ref{starsample}.

244: 

245: The data taken with slit spectrographs, i.e., all except the observations

246: with HYDRA@CTIO and WYFFOS@WHT, were reduced following the procedure

247: described by \citet{massey92} using the IRAF\footnote{IRAF is

248: distributed by the National Optical Astronomy Observatory, which is

249: operated by the Association of Universities for Research in Astronomy,

250: Inc., under cooperative agreement with the National Science

251: Foundation.} packages but with some small differences described by

252: \citet{pont04}.  We obtained two images of each object, with the star

253: shifted along the slit. First, we subtracted the bias and overscan,

254: and corrected by the flat-field. Then, since the star is in a

255: different physical position in the two images, we subtracted one from

256: the other, obtaining a positive and a negative spectrum in the same

257: image.  With this procedure the sky is subtracted in the same physical

258: pixel in which the star was observed, thus minimizing the effects of

259: pixel to pixel sensitivity variations. Of course, a time dependency

260: remains since the two spectra have not been taken

261: simultaneously. These sky residues are eliminated in the following

262: step, when the spectrum is extracted in the traditional way and the

263: remaining sky background is subtracted from the information on both

264: sides of the star aperture. As the next step, the spectrum is

265: wavelength calibrated.  We then again subtracted the negative from the

266: positive (so we added both spectra because one is negative) to obtain

267: the final spectrum. Finally, each spectrum was normalized by fitting a

268: polynomial, excluding the strongest lines in the wavelength range such

269: as those of the CaT. The order of the polynomial changes among runs in

270: order to eliminate the response of each instrument. The wavelength

271: calibration of the VLT data (both from the archive and from run 6)

272: might be less accurate than the rest because arcs are not taken at the

273: same time and with the same telescope pointing as the object. The

274: effects of this on the wavelength calibration has discussed by

275: \citet{gallart01}, and we evaluate them in Section

276: \ref{radialvelocities}. However, since we are not interested in

277: obtaining precise radial velocities, this problem will not have an

278: important impact on our project.

279: 

280: HYDRA@CTIO and WYFFOS@WHT are multifibre spectrographs. The data

281: obtained with HYDRA has been extracted with the DOHYDRA task within

282: IRAF in the way described by \citet{valdes92}.  This task was

283: developed specially to extract data acquired with this instrument. The

284: procedure is described in depth by \citet{carrera05}. Basically, after

285: bias, overscan subtraction and trimming, DOHYDRA traces the apertures,

286: makes the flat-field correction and calibrates in wavelength. We

287: followed a similar procedure with the data obtained with WYFFOS, but

288: in this case we used the general DOFIBERS task, which works similarly

289: to DOHYDRA. Although both tasks allow for sky subtraction, the results

290: were poor, and important residuals of sky lines remained. To remove

291: the contribution of these sky lines, we have developed our own

292: procedure to subtract them. Basically, it consists in obtaining an

293: average sky spectrum from all fibres placed on the sky in a given

294: configuration. Before subtracting this average, high S/N sky, from

295: each star spectrum, we need to know the relation between the intensity

296: of the sky in each fibre (which varies from fibre to fibre due to the

297: different fibre responses) and the average sky. This relation is a

298: weight (which may depend on wavelength) by which we must multiply the

299: average sky spectra before subtracting it from each star. To calculate

300: it, we have developed a task that finds the weight which minimizes the

301: sky line residuals over the whole spectral region considered. As a

302: result of this procedure, the sky emission lines are removed very

303: accurately. Finally, the normalization was carried out in the same way

304: as previously described.

305: 

306: Examples of 4 stars with different metallicities are shown in Figure

307: \ref{spectra}. Note how the strength of the CaT lines increases with

308: metallicity.

309: 

310: \placefigure{spectra}

311: 

312: %\subsection{Radial Velocities\label{radialvelocities}}

313: 

314: The radial velocity of each star has been calculated in order to

315: reject cluster non-members. We used the FXCOR task in IRAF, which

316: performs the cross-correlation between the target and template spectra

317: of known radial velocity \citep{td79}. We selected between 8 and 10

318: template stars in each run that had very high S/N and covered a wide

319: range of radial velocities. The velocities were corrected to the

320: heliocentric reference frame within FXCOR. The final radial velocity

321: for each star was obtained as the average of the velocities obtained

322: from each template, weighted by the width of correlation peaks.

323: 

324: In the case of observations with slit spectrographers, the star might

325: not be exactly positioned in the centre of the slit. This error means

326: a velocity uncertainty given by $\Delta v=c\times \Delta \Theta \times

327: p/\lambda_0$, where: $c$ is the light speed, $p$ is the spectral

328: resolution given in \AA\ arcsec$^{-1}$; $\lambda_0$ is the wavelength

329: of the lines (in this case $\sim$8600 \AA), and $\Delta \Theta$ is the

330: angular offset of the star from the centre of the slit in arcsec. This

331: effect has been described by \citet{irwint02} and \citet{hz06}. In our

332: case, it may only be significant in the case of the VLT

333: observations. To estimate the offset in this case we used through-slit

334: images obtained at the beginning of the observation of each

335: configuration, taken to check that the stars were positioned in the

336: slits.  In this image we have measured the position of each stellar

337: centroid, which is compared with the position of the slit given in the

338: header of the image. The difference between both, $\Delta \Theta$,

339: allows us to calculate the uncertainty in the measurement of the

340: radial velocity. This value changes from one star to another, the

341: error being about 15 km s$^{-1}$ on average.

342: 

343: The mean velocity for each cluster is listed in Table

344: \ref{samplevr}. Most of the values obtained agree, within the

345: uncertainties, with previous measurements from the literature, even in

346: the case of the clusters observed with the VLT, where the

347: uncertainties are larger. In the case of NGC 2141, we found a mean

348: velocity similar to the value obtained by \citet{c04}. Both values

349: differ by 20 and 30 km s$^{-1}$, respectively, from the value found by

350: \citet{f02}. For Collinder 110, no previous measurement of its radial

351: velocity could be found in the literature.

352: 

353: \section{The Calcium Triplet \label{catriplet}}

354: 

355: We are interested in obtaining metallicities from red giant stars, and

356: within this group, from the brightest ones, which are of spectral

357: types K and M. The main features in the infrared spectra of these

358: stars are the CaT lines. But their spectra also contains other weak

359: atomic lines. The Fe {\footnotesize I} (8514.1, 8674.8, 8688.6 and

360: 8824.2 \AA) and Ti {\footnotesize I} (8435.0 \AA) lines are the most

361: important. When within this range, we move to later spectral types,

362: and hence to cooler stars, molecular bands begin to appear that change

363: the slope of the local continuum.  The main contribution are from the

364: titanium oxide (TiO) bands, the strongest of which are the triplet

365: situated at 8432, 8442 and 8452 \AA~and the doublet at 8859.6 and

366: 8868.5 \AA. There are other weaker bands at 8472, 8506, 8513, 8558 and

367: 8569 \AA, near the bluest lines of the CaT. There are also several

368: vanadium oxide (VO) bands at 8521, 8538, 8574, 8597, 8605, 8624, 8649

369: and 8668 \AA. The strength of these features increases when the

370: temperature decreases, i.e.\ when we move to later spectral types. The

371: presence of these bands complicates the definition of the continuum,

372: which makes it difficult to obtain the equivalent widths of the CaT

373: lines for stars with T$_{eff}\leq$3500 K or (V-I)$>$2, in the most

374: metal-rich clusters. The description of the CaT region for other

375: spectral types can be found in \citet{cen01}.

376:    

377: \subsection{Definition of Line and Continuum Bandpass Windows}

378: 

379: In the literature we can find different prescriptions to measure the

380: strength of the CaT lines. The classical definition of a spectral

381: index consists in establishing a central bandpass covering a spectral

382: feature and one or more bandpasses on both sides to trace the local

383: continuum reference level. \citet{cen01} have presented a description

384: of the previous CaT index definitions and a comparison among them. In

385: Figure \ref{bandas} we have plotted the line and continuum bandpasses

386: used in several reference works, \citet{cen01} (a), \citet{r97a} (b)

387: and \citet{az88} (c), over a metal-poor (left) and a metal-rich

388: (right) spectrum. The \citet{az88} and \citet{r97a} indices were

389: defined for relatively metal-poor RGB stars where the influence of the

390: molecular bands is not important. The index of \citet{cen01} was

391: defined specifically to avoid the presence of molecular bands. Also,

392: from Figure \ref{bandas}, we can easily see that the wings of the

393: lines are larger than the line bandpasses defined by \citet{az88} and

394: \citet{r97a} in the case of the metal rich stars. Only the line

395: bandpasses defined by \citet{cen01} completely cover the line

396: wings. Although we have selected the bandpasses defined by

397: \citet{cen01}, which are listed in Table

398: \ref{bandastable}, the equivalent width of the line will be measured

399: in a different way, as described in the following section.

400: \placefigure{bandas}

401: 

402: \subsection{Equivalent widths\label{ew}}

403: 

404: The next step is to measure the line flux from its equivalent

405: width. The equivalent width of a spectral line can be measured in

406: different ways. One method is by numerical integration of the observed

407: spectra in a line band \citep[e.g.][]{cen01}. However, in the wings of

408: the strongest lines of the CaT there are some weak lines, whose

409: strength may change with different stellar atmospheric parameters

410: than the CaT lines. These lines must be excluded when we measure the

411: CaT equivalent width. The alternative \citep[e.g.][]{r97a,c04}

412: consists in fitting an empirical function to a line profile and

413: calculating the equivalent width from the integration of this fit.

414: Many functions have been used to fit the CaT line profiles, most

415: commonly a Gaussian profile \citep[e.g.][]{adc91}. However, as

416: \citet{c04} have shown, the Gaussian profile provides a good fit for

417: weak-line stars, but the fit is worse in strong-line stars, where the

418: contribution of the non-Gaussian wings of the CaT lines becomes

419: substantial. We have to take this point into account because the main

420: contributors to the strength of the CaT lines are their wings, while

421: the core is not very sensitive to the atmosphere and stellar

422: parameters \citep{erdelyi91}. \citet{r97a} fitted a Moffat function of

423: exponent 2.5. As \citet{pont04} has demonstrated, the behaviour of

424: Moffat function of exponent 2.5 is similar to the Gaussian fit for the

425: weakest lines. However, neither provides a good fit to the strongest

426: lines. \citet{c04} fitted the whole line profile with the sum of a

427: Gaussian and a Lorentzian function, which provides a better fit for

428: the strongest lines and agrees with the single Gaussian fit for the

429: weakest lines \citep[see][for a further discussion]{c04}. We have

430: compared the different functions in order to evaluate the quality of

431: the fit in the whole range of line strengths. We have chosen the sum

432: of a Gaussian and a Lorentzian function because this provides the best

433: fit for the whole range of equivalent widths in this study. We have

434: also checked whether a simple Gaussian or Moffat function would

435: produce a good fit in the case of spectra obtained with lower

436: resolution. Also in this case a Gaussian plus a Lorentzian provides

437: the best fit for strong-line stars.

438: 

439: A Gaussian plus a Lorentzian function has therefore been fitted to the

440: line profiles with a least-squares method, using the

441: Levenberg-Marquardt algorithm. For the whole range of equivalent

442: widths covered in this work, the differences between the observed line

443: and the fit are negligible for stars with S/N $\geq$ 20. Stars

444: with poorer S/N have been rejected. The equivalent width of each

445: line is the area limited by the fitted profile of the line and the

446: continuum level, defined as the linear fit to the mean values of the

447: flux in each window chosen to determine the continuum. Formal errors

448: of the fit are estimated as the difference between the equivalent

449: width measurement for continuum displacements of $\pm(S/N)^{-1}$.

450: 

451: \subsection{The CaT index \label{catindex}}

452: 

453: The equivalent widths of the three CaT lines are combined to form the

454: global index $\Sigma$Ca \citep{adc91}. Some authors excluded the

455: weakest line at 8498 \AA\ on the basis of its poor S/N

456: \citep[e.g.][]{sunt93, c00}. Others have used all three lines, either

457: weighted \citep[e.g.][]{r97a} or unweighted \citep[e.g.][]{ol91}. As

458: our spectra have high S/N ratios, we used the unweighted sum of the

459: three lines, $\Sigma Ca$=W$_{8498}$+W$_{8542}$+W$_{8662}$, and

460: we calculate its error as the square root of the quadratic sum of the

461: errors of each line. As we have some stars in common with previous

462: works, we can compare the $\Sigma Ca$ calculated by us with values

463: obtained in previous papers. \citet{r97a} compared their $\Sigma$Ca

464: with previous index definitions until 1997. Here, for simplicity, we

465: are only going to compare our index with three reference works. Stars

466: in common with \citet{adc91, r97a} and \citet{c04} are plotted in

467: Figure \ref{comp}. As mentioned before, the works of \citet{adc91} and

468: \citet{r97a} were focused on old and metal-poor stars. However,

469: \citet{ol91} and \citet{sunt93}, using the same index as \citet{adc91}  

470: defined for globular cluster stars, measured the equivalent width of

471: the CaT lines in stars of two open clusters, M11 and M67,

472: respectively. We are going to use these values to complete the

473: measurements of \citet{adc91}.

474: 

475: \placefigure{comp}

476: 

477: We find a quasilinear relation up to $\Sigma Ca\sim$7 among the

478: $\Sigma Ca$ values in this paper and those obtained by

479: \citet{adc91} \citep[see also][]{sunt93}. From this point the relationship

480: saturates: while our index increases by an additional $\Delta\Sigma

481: Ca\sim$2, theirs only increases by $\Delta\Sigma Ca\sim$1.5 (on their

482: scale). We believe that the reason for this is that they fitted the

483: line profile by a Gaussian function which underestimates the

484: contribution of the line wings in strong lines (see Section~

485: \label{catindex}).  Note also the zero-point difference between both

486: scales. The relation is not exactly one to one because they did not

487: use the equivalent width of the weakest CaT line. However, the slope

488: close to one of the linear fit for the metal-poor stars implies that

489: the two indices are almost equivalent for these kind of stars. The

490: loss of linearity for strong-line stars partly explains why these

491: authors found a nonlinear relationship between the CaT index and

492: metallicity, but, of course, the metallicity scale also plays a role

493: in this issue, as we discuss in Section \ref{othermetallicities}. The

494: linear fit for $\Sigma$Ca$\leq$7 (solid straight line in top panel of

495: Figure

496: \ref{comp}) is:\\

497: 

498: \begin{equation}

499: \Sigma Ca_{AC91}=-0.88(\pm0.08)+0.96(\pm0.01)\Sigma Ca_{TP}

500: \end{equation}

501: 

502: and the second order polynomial fit for the whole range of equivalent widths is\\

503: 

504: \begin{equation}

505: \Sigma Ca_{AC91}=-1.10(\pm0.08)+1.20(\pm0.03)\Sigma Ca_{TP}-0.04(\pm0.01)\Sigma Ca_{TP}^2

506: \end{equation}

507: 

508: In the case of \citet{r97a}, who only observed stars with [Fe/H]

509: $\leq-0.7$, we find a linear correlation for the whole range of

510: equivalent widths. In this case the slope is less than one, meaning

511: that their index is less sensitive to changes in the strength of the

512: CaT lines than ours. For the same star, our index is higher than the

513: \citet{r97a} one. The linear fit is:\\

514: 

515: \begin{equation}

516: \Sigma Ca_{R97}=-0.23(\pm0.06)+0.78(\pm0.01)\Sigma Ca_{TP}.

517: \end{equation}

518: 

519: Finally, the correlation between \citet{c04} index and ours is one to

520: one ($\Sigma Ca_{TP}-\Sigma Ca_{C04}=0.009\pm0.0007$). As we used the

521: same empirical function and index definition of $\Sigma$Ca as

522: \citet{c04}, differences could only come from the definition of line

523: and continuum bandpasses. This means that, in the range of equivalent

524: widths covered here, both indices are equivalent. However, as the

525: continuum in our index has been defined to avoid the influence of TiO

526: bands, we expect that our index would also behave well in stars whose

527: continuum is contaminated by TiO bands.

528: 

529: \subsection{The reduced equivalent width\label{reducedew}}

530: 

531: \placefigure{luminosityfig}

532: 

533: The next step is to relate the CaT index with metallicity. The

534: strength of the absorption lines mainly depends on the chemical

535: abundance, stellar effective temperature (T$_{eff}$) and surface

536: gravity (log $g$). Therefore, to relate the equivalent width of the

537: CaT lines with metallicity it is necessary to remove the T$_{eff}$ and

538: log $g$ dependence. \citet{adc91} and \citet{ol91} demonstrated that

539: the cluster stars define a sequence in the Luminosity--$\Sigma Ca$

540: plane, using luminosity measures from indicators like M$_I$ or

541: (V-V$_{HB}$). These sequences are separated as a function of the

542: cluster metallicity. The theoretical explanation of this can be found

543: in \citet{pont04}, using \citet{jcj92} models, which describe the

544: behaviour of the CaT lines as a function of T$_{eff}$, log $g$ and

545: metallicity.

546: 

547: It is necessary to study the morphology of the sequence defined by

548: each cluster in the Luminosity--$\Sigma Ca$ plane. From a theoretical

549: point of view, the increment of luminosity along the RGB comes with a

550: drop in T$_{eff}$ and log $g$ that decreases and increases the

551: strength of the lines, respectively. The result is a modest increment

552: in $\Sigma Ca$ with luminosity ($\delta\Sigma Ca/\delta

553: M_I\sim$0.5). Moreover, the models predict that $\Sigma Ca$ increases

554: more rapidly with luminosity in the upper part of the RGB (above the

555: HB) than in the lower part.  In other words, the sequence defined by

556: each cluster might not be linear and might be best described adding a

557: quadratic component. The \citet{jcj92} models also predict that

558: $\Sigma Ca$ increases more rapidly when log $g$ decreases, or when the

559: luminosity increases, for the more metal-rich clusters than for the

560: more metal-poor ones. Therefore, the linear and quadratic terms, which

561: characterize the sequence defined for each cluster in the

562: luminosity--$\Sigma Ca$ plane, increase with metallicity, as can be

563: seen in Figure 15 of \citet{pont04}.

564: 

565: Observationally, the variation in $\Sigma Ca$ with metallicity has

566: traditionally been studied from (V-V$_{HB}$), which removes any

567: dependence on distance and reddening

568: \citep[e.g.][]{adc91,r97a,c04}. In this context, it is found that

569: clusters define linear sequences in the (V-V$_{HB}$)--$\Sigma Ca$

570: plane, where the reduced equivalent width, $W'$, is defined as $\Sigma

571: Ca=W'_{HB}$+$\beta$(V-V$_{HB}$). \citet{r97a} found that the slopes

572: of these sequences were the same for all clusters in their sample,

573: independently of their metallicity. Therefore only $W'_{HB}$ changes

574: from one cluster to another, and its variation is directly related to

575: metallicity. Other studies have reached the same conclusion using open

576: and globular clusters \citep[e.g.][]{ol91}. \citet{pont04} \citep[see

577: also][]{adc91} have demonstrated that this also occurs in the

578: M$_V$-$\Sigma Ca$ and M$_I$-$\Sigma Ca$ planes. However, no studies

579: have observed the theoretical predictions that cluster sequences are

580: not exactly linear with luminosity, or that their shape depends on

581: metallicity.

582: 

583: The main objective of this study is to apply the relationships

584: obtained to derive metallicities of individual stars in Local Group

585: galaxies, which in general have had multiple star formation epochs and

586: do not always have a well defined HB (e.g. LMC: Carrera et al. 2007;

587: SMC: No\"el et al. 2007; Leo A: Cole et al. 2007). For example, the

588: Magellanic Clouds do not have a measurable HB in the CMD, and in

589: studies which define the reduced equivalent width as a function of

590: $(V-V_{HB})$ (\citep[e.g.][]{c05}), the HB position has been taken as

591: that of the red-clump. However, in the Magellanic Clouds, the position

592: of the red-clump is about 0.4 magnitudes brighter than the HB. This

593: only implies underestimating the metallicity by $\simeq$ 0.15 dex,

594: which is similar to the uncertainty on the metallicity determination

595: itself. Distances to Local Group galaxies are in general determined

596: with an accuracy greater than 0.4 mag., and so, even if the error on

597: the derived metallicity due to the uncertainty in the position of the

598: HB is not large, it can be minimized by defining the reduced equivalent

599: width as a function of absolute magnitude. This point is also

600: important in the case of open clusters, which hardly ever have a

601: HB or, if they do, it is usually not well defined.  For

602: this reason, like \citet{pont04}, we redefine $W'$ as the value of

603: $\Sigma Ca$ at M$_V$=0 (hereafter $W'_V$) or M$_I$=0 (hereafter

604: $W'_I$).

605: 

606: First we will study in detail the morphology of the cluster sequences

607: in the Luminosity--$\Sigma Ca$ plane. As discussed above, from a

608: theoretical point of view, we expect that these sequences are not

609: exactly linear. We have observed stars along the RGB in 5 clusters

610: covering the whole metallicity range. In Figure \ref{luminosityfig} we

611: have plotted stars observed in these clusters in the M$_V$--$\Sigma

612: Ca$ and M$_I$--$\Sigma Ca$ planes. These stars have magnitudes in the

613: ranges -2$\leq$M$_V\leq$2 and -3$\leq$M$_I\leq$2 (or

614: -2.3$\leq$V-V$_{HB}\leq$1.8). These ranges contain both stars brighter

615: and fainter than previous works \citep[e.g.][]{r97a,c04}. Note that

616: the strength of the CaT lines increases more rapidly in the upper part

617: of the RGB, as predicted by \citet{pont04} using \citet{jcj92}

618: models. These observations can be used to obtain a new relationship

619: between $\Sigma Ca$, absolute magnitude and metallicity valid for all

620: the stars in the RGB, that takes into account the curvature in the

621: Luminosity--$\Sigma Ca$ plane.  The sequence of each cluster has been

622: fitted with a quadratic function such that $\Sigma

623: Ca$=$W'_{V,RGB}$+$\beta$M$_V$+$\gamma$M$_V^2$. We plotted the result

624: when the stars of each cluster are fitted independently in

625: Figure~\ref{luminosityfig}. The coefficients of the fit are shown in

626: Table \ref{luminositytable}. From this, it seems that $\beta$ tends to

627: increase with metallicity, as predicted theoretically. In the case of

628: $\gamma$ this increment is not observed, i.e.  its variation does not

629: show a significant dependence on metallicity, except for the most

630: metal-rich cluster, which also has a large uncertainty.

631: 

632: Using the \citet{jcj92} empirical relations and the BaSTI stellar

633: evolution models \citep{pie04}, we have calculated theoretical

634: sequences for clusters with [Fe/H] $\geq -1$, which are plotted in

635: Figure \ref{luminositymodels} as dashed lines. These models were

636: obtained for [Fe/H] = +0.5, 0, $-0.5$ and $-1$, while the clusters

637: metallicities are [Fe/H] =+0.47, $-$0.14, $-$0.67 and $-$1.07

638: respectively. \citet{jcj92} did not compute relationships for more

639: metal-poor clusters. We used BaSTI isochrones with metallicities of

640: +0.32, $-0.28$, $-$0.58 and $-$0.98, respectively, in order to

641: estimate T$_{eff}$ and log $g$ along the RGB. The \citet{jcj92}

642: relationships were calculated for the two strongest CaT lines. To

643: compare the theoretical predictions with the observational sequences

644: we computed, using our own data, an empirical relation between $\Sigma

645: Ca_{8442+8662}$ obtained from these two lines and the $\Sigma Ca$ used

646: in this work, computed from the three CaT lines. We found is $\Sigma

647: Ca=0.13+1.21\Sigma Ca_{8442+8662}$. Applying this correction, we find

648: that the theoretical and observed cluster sequences still do not

649: match. There is a zero-point that changes from one cluster to another,

650: which is not surprising because the cluster metallicities are

651: not exactly the same as those used to compute the theoretical

652: relationships. Therefore, the theoretical sequences have been

653: shifted in order to superimpose them on the cluster ones. It can be

654: seen that models do not exactly reproduce the behaviour of the

655: observed cluster sequences. However, the prediction that the shape

656: changes from the metal-poor clusters to the metal-rich ones is

657: observed, although, as was mentioned before, these variations are

658: similar to the uncertainties.

659: 

660: We can simplify the problem if we assume that all clusters have the

661: same tendency, i.e. if we calculate a single slope and quadratic term

662: for the whole sample. So only the zero point changes among

663: clusters. To obtain these coefficients, we have performed an iterative

664: least-squares fit as described by \citet{r97a}. From a set of

665: reference values, we obtained the quadratic and linear terms of the

666: fit in iterative steps, until they converged to a single value within

667: the errors and allow only the zero point to change among clusters. The

668: values are: $\beta_V=-0.647\pm 0.005$ and $\gamma_V=0.085\pm0.006$. In

669: the same way, for M$_I$ we obtained $\beta_I=-0.618\pm 0.005$ and

670: $\gamma_I=0.046\pm 0.001$. In Figure \ref{luminositymodels} we have

671: plotted the individual fit for each cluster (solid line) and that when

672: the linear and quadratic terms do not change among clusters (dashed

673: lines). In both cases, the dotted lines represent the region where

674: there are no cluster stars and the fits have therefore been

675: extrapolated. As we can see in Figure \ref{luminositymodels}, in the

676: magnitude interval covered by cluster stars, both fits are similar and

677: give very similar values of $W'$ within the uncertainties. For

678: example, for NGC 7078, where the discrepancy is larger, we obtained

679: 2.79 $\pm$ 0.06 and 2.79 $\pm$ 0.01 in $V$; and 2.64 $\pm$ 0.08 and

680: 2.31 $\pm$ 0.01 in $I$, when the linear and quadratic terms change

681: among clusters or they are fixed, respectively. Larger differences

682: between both fits are found in the regions where the relationships are

683: extrapolated.

684: %where the condition M$_V$=0 (or M$_I$=0) hardly ever obtains and

685: 

686: Moreover, in our case we are interested in measuring the strength of

687: the CaT lines in galaxies where we can observe only the upper part of

688: the RGB with a good S/N. The quadratic behaviour of the cluster

689: sequences in the Luminosity--$\Sigma Ca$ plane is not significant when

690: we observe stars with M$_I\leq$0 only (or M$_V\leq$1.25; this

691: magnitude limit has been selected in order to sample in both filters

692: the same number of stars in each cluster). For example, when we repeat

693: the previous procedure, but only for stars with M$_V\leq$ 1.25, we

694: find that the quadratic term is $\gamma_V$ = 0.004 $\pm$ 0.003, which

695: is negligible within the uncertainty. In the same way, when we only

696: observe stars with M$_V\geq$ 1.25 we obtain a similar result:

697: $\gamma_V=0.002\pm 0.01$. The same happens in the M$_I$--$\Sigma Ca$

698: plane, but here the quadratic terms are even smaller. According to

699: this, the cluster sequence can be considered linear above and below

700: M$_V$=1.25 and M$_I$=0, and we can fit it as $\Sigma$Ca =

701: $W'_V+\beta_V M_V$ or $\Sigma$Ca = $W'_I+\beta_I M_I$ on each side of

702: this point. Following the same iterative procedure as in the case of

703: the quadratic fit, we calculated the values of the slope $\beta$ for

704: M$_V\leq$ 1.25 and for M$_I\leq$ 0, obtaining $\beta_{V}=-0.74\pm0.01$

705: and $\beta_{I}=-0.60\pm0.01$, respectively. The linear fits for

706: M$_V\leq$1.25 and M$_I\leq$ 0 are represented in Figure

707: \ref{luminositymodels}, by dotted--dashed lines. In all cases, within

708: the ranges covered by the cluster stars, the linear fit to the bright

709: stars is equivalent, within the uncertainties, to the quadratic ones.

710: 

711: Finally, for clusters where we have observed a wide range of

712: magnitudes we find that the slope ($\beta$) increases, although within

713: the uncertainties, with metallicity. We might check this point using

714: now all clusters in our sample. A total of 27 clusters in $I$ and 29

715: in $V$ have stars brighter than M$_I$=0 and M$_V$ = 1.25. We have

716: fitted the sequence to each cluster independently in the linear form

717: $\Sigma Ca=W'_{V,I}+\beta_{V,I} M_{V,I}$. The values obtained from the

718: slope have been plotted against $W'$, which is directly correlated with

719: metallicity, for each cluster in Figure~\ref{pendientefig}.  From this

720: figure it is seen that there is no significant relation between the

721: cluster slope and $W'$ (or [Fe/H]).  Therefore, from here on, we consider the

722: slope of the fit to be the same for the whole range of [Fe/H] and,

723: hence, for all objects.

724: 

725: In summary, as we are specially interested in obtaining metallicities

726: for stars in the upper part of the RGB with the CaT, where the

727: quadratic term is not significant and the slope can be fixed

728: independently of metallicity, we are going to use a linear fit with a

729: single slope for the calibration using the whole cluster sample. This

730: is what has been done in all previous calibrations of the CaT.

731: 

732: Figures \ref{mvsigmaca} and \ref{misigmaca} represent the clusters in

733: our sample in the M$_V$--$\Sigma Ca$ and M$_I$--$\Sigma Ca$ planes

734: respectively, together with the linear fit to each of them. Using the

735: same procedure as in the case of the quadratic fit discussed above, we

736: have obtained $\beta_V=-0.677\pm0.004$ and

737: $\beta_I=-0.611\pm0.002$\AA~mag$^{-1}$ in the M$_V$--$\Sigma Ca$ and

738: M$_I$--$\Sigma Ca$ planes, respectively. The value found in the

739: M$_I$--$\Sigma Ca$ plane is slightly larger than that obtained by

740: \citet{pont04}, $\beta_I=-0.48\pm0.02$ \AA~mag$^{-1}$. Although these 

741: authors used a different method to calculate the metallicity (they

742: fitted each cluster individually and obtained the mean of the slopes

743: of all of them), this is not the reason for the discrepancy because if

744: we follow the same procedure with our own data, again we find

745: $\beta_I=-0.61$.  There are no previous determinations of

746: $\beta_V$. The values obtained for $W'_V$ and $W'_I$ are listed in

747: Table \ref{fitstable}.

748:  

749: \placefigure{luminositymodels}

750: 

751: \placefigure{mvsigmaca}

752: 

753: \placefigure{misigmaca}

754: 

755: \section{The Ca {\footnotesize II} Triplet metallicity scale\label{catmetallicityscale}}

756: 

757: An important point in this study is the reference metallicities. It

758: would be ideal to use the same metallicity scale for both open and

759: globular clusters, and that this would have been obtained from

760: high-resolution spectroscopy. In the literature we can find two

761: globular cluster metallicity scales obtained from high resolution

762: spectroscopy: \citet[hereafter CG97]{cg97} and \citet[hereafter

763: KI03]{ki03}. There is a third metallicity scale obtained from

764: low-resolution data: \citet[hereafter ZW84]{zw84}. There are

765: systematic differences among these three scales, but there is no

766: reason to prefer any particular one of them. For this reason, here we

767: are going to study the behaviour of the CaT lines with metallicity in

768: these three scales. Lamentably, there is not a homogeneous metallicity

769: scale obtained from high-resolution spectroscopy for open

770: clusters. However, the metallicities of some of them have been

771: obtained directly in the CG97 scale by some authors: NGC 6819

772: \citep{bra01}; NGC 2506 \citep{cbgt04}; NGC 6791 \citep{gratton06} and 

773: Berkeley 32 \citep{sestito06}. These metallicities were obtained using

774: Fe {\footnotesize I} and Fe {\footnotesize II} lines. For the other 8

775: open clusters in our sample there are also metallicities obtained from

776: high-resolution spectroscopy in RGB stars and using Fe

777: {\footnotesize I} and Fe {\footnotesize II} lines in a similar way to

778: CG97. Even though some discrepancies could exist because the

779: procedures are not exactly the same, we are considering these

780: metallicities also to be on the CG97 scale. The reference values in

781: this scale are listed in column 2 of Table \ref{clustersample} and the

782: sources for each of them are listed in column 3. The reference

783: metallicities in the ZW84 and KI03 are listed in columns 4 and 5

784: respectively. In both cases, we have used only values obtained

785: directly by these authors.

786: 

787: \subsection{Calibration in the CG97 metallicity scale\label{cg97calibration}}

788: 

789: \placefigure{calv}

790: 

791: Figures \ref{calv} and \ref{calI} show the run of $W'_V$ and $W'_I$

792: with metallicity. In most cases, the errors are smaller than the size

793: of the points. The circles indicate clusters younger than 4 Gyr. The

794: solid line shows the best fit to the data. The dashed lines represent

795: the 90\% confidence level. Note that in both cases there is a linear

796: correlation. The bottom panels show the residuals of the linear

797: fit. We have used 22 clusters for the calibration in $V$ and 20 for

798: that in $I$. There are three clusters that differ from the fit by more

799: than 0.2 dex in both filters. These clusters are NGC 2420, NGC 2506

800: and Berkeley 32. They have been excluded from the analysis. In the

801: case of NGC 2420, only 6 stars in $V$ and 4 in $I$ are radial velocity

802: members. This, together with a relatively large uncertainty in its

803: metallicity \citep{gratton00}, contributes to its large error bar. In

804: the case of NGC 2506 and Berkeley 32, there are only 3 and 4 stars

805: respectively with membership confirmed by their radial

806: velocities. Thus, slight differences in the $\Sigma Ca$ value of one

807: of them could change the derived $W'$ significantly. Two of the three very deviant clusters (NGC 2420 and NGC 2506) have ages

808: less than 4 Gyrs, but 5 other young clusters fit the mean relationships in

809: Figures \ref{calv} and \ref{calI} to better than 0.2 dex. We doubt therefore that cluster

810: age is the major cause of the large deviations. 

811: 

812: \placefigure{calI}

813: 

814: The best linear fits shown in Figures \ref{calv} and \ref{calI}, are:

815: 

816: \begin{equation}

817: [Fe/H]_{CG97}^V=-3.12(\pm0.06)+0.36(\pm0.01)W'_V~~\sigma_V=0.08\label{cg97v}

818: \end{equation} 

819: 

820: \begin{equation}

821: [Fe/H]_{CG97}^I=-2.95(\pm0.06)+0.38(\pm0.01)W'_I~~\sigma_I=0.09\label{cg97i}

822: \end{equation} 

823: 

824: Some studies have predicted that this relationship may present a

825: curvature due to the loss of CaT index sensitivity at high

826: metallicities \citep[e.g.][]{dtt89}. \citet{c04} investigated this

827: point adding a quadratic term. They found that the coefficient of this

828: term is insignificant and does not improve the quality of the fit. We

829: performed the same analysis in our sample, which covers a wider range

830: of ages and metallicities, finding a similarly insignificant influence

831: of a quadratic term.

832: 

833: \subsection{Calibration on Other Metallicity Scales\label{othermetallicities}}

834: 

835: In this section we study the behaviour of the CaT on the ZW84 and KI03

836: scales. In Figure \ref{metallicities} we have plotted the

837: metallicities in ZW84 (bottom) and KI03 (top) listed in Table

838: \ref{clustersample} versus $W'_V$ (left) and $W'_I$ (right), respectively. 

839: 

840: In the case of the KI03 metallicity scale (top panels), the behaviour

841: of $W'$ with metallicity is linear, as for the CG97 scale. These

842: authors used three stellar atmosphere models to obtain

843: metallicities. For simplicity, in Figure \ref{metallicities} we have

844: plotted only the metallicity values obtained using MARCS

845: models. However, a linear behaviour is also found when we use the

846: metallicities computed from the Kurucz models with or without

847: convective overshooting. The linear fits for each of the three models

848: are:

849: 

850: \begin{mathletters}

851: \scriptsize

852: \begin{eqnarray}

853: &[Fe/H]_{KI03}^V&=-3.42(\pm0.03)+0.37(\pm0.01)W'_V~\sigma=0.10~(MARCS) \label{ki03vm}\\

854: &[Fe/H]_{KI03}^V&=-3.43(\pm0.03)+0.38(\pm0.01)W'_V~\sigma=0.10~(Kurucz~with~convective~overshooting) \\

855: &[Fe/H]_{KI03}^V&=-3.51(\pm0.03)+0.40(\pm0.01)W'_V~\sigma=0.10~(Kurucz~without~convective~overshooting) 

856: \end{eqnarray}

857: \end{mathletters}

858: 

859: \begin{mathletters}

860: \scriptsize

861: \begin{eqnarray}

862: &[Fe/H]_{KI03}^I&=-3.29(\pm0.03)+0.40(\pm0.01)W'_I~\sigma=0.09~(MARCS)\label{ki03im} \\

863: &[Fe/H]_{KI03}^I&=-3.24(\pm0.03)+0.40(\pm0.01)W'_I~\sigma=0.09~(Kurucz~with~convective~overshooting) \\

864: &[Fe/H]_{KI03}^I&=-3.31(\pm0.03)+0.41(\pm0.01)W'_I~\sigma=0.09~(Kurucz~without~convective~overshooting)

865: \end{eqnarray} 

866: \end{mathletters}

867: 

868: Differences between metallicities derived with the MARCS

869: model and the models of Kurucz with or without overshooting are negligible.

870: 

871: This linear behaviour is not surprising because, as KI03 demonstrated,

872: their metallicities are linearly correlated with the CG97 values,

873: which are, at the same time, linearly correlated with our

874: $W'$. However, the metallicities calculated by KI03 are systematically

875: lower than the CG97 ones. KI03 studied this point and concluded that

876: the difference could be explained because they used different

877: T$_{eff}$ and log g values, as well as different atmosphere

878: models. The combination of all these can easily introduce systematic

879: differences in the globular cluster abundance scales.

880: 

881: In the case of ZW84, we have found that the data

882: are best fitted by a second-degree polynomial (solid line):

883: 

884: \begin{mathletters}

885: \begin{eqnarray}

886: &[Fe/H]_{ZW84}^V&=-1.98(\pm0.07)-0.18(\pm0.02)W'_V+0.05(\pm0.01)W'^2_V~~\sigma_V=0.10\\

887: &[Fe/H]_{ZW84}^I&=-2.07(\pm0.07)-0.12(\pm0.03)W'_I+0.05(\pm0.01)W'^2_I~~\sigma_I=0.09\label{zw84i}

888: \end{eqnarray} 

889: \end{mathletters}

890:  

891: In Section 4.3, we discussed several previous definitions and

892: measurement procedures of the CaT lines, and noted the loss of

893: sensitivity to the CaT lines strength in some cases (e.g. Armandroff

894: and Da Costa 1991) which also found a non-linear relationship between

895: the CaT index and metallicity. We mentioned that this non-linearity

896: was probably the result of the combination of a non-accurate

897: measurement of the CaT on strong-line stars and the particular

898: metallicity scale in use. In order to assess the relative importance

899: each factor, we will now compare the effects on the derived abundances

900: of alternatively i) assuming a linear relationship between $W'$ and

901: metallicity on the ZW84 metallicity scale and ii) adopting a Gaussian

902: to fit the CaT lines, which provides a poorer fit. When a linear

903: relationship between $W'_I$ and [Fe/H]$_{ZW84}$ is assumed, the derived

904: metallicity of a strong-line star, $W'_I$=8.5, is underestimated in

905: 0.3 dex. In the case of a weak-line star, $W'_I$=2, again the

906: metallicity is underestimated in 0.2 dex. Similar results are obtained

907: when lines are not properly fitted. For example, as we saw in Section

908: \ref{catindex}, \citet{adc91} fitted the line profile with a Gaussian, 

909: resulting in that their index saturated for strong-line stars. The

910: relation between the reduced equivalent width obtained from their

911: index and metallicities in the CG97 scale is a second-degree

912: polynomial. If we then assume a linear relationship between this index

913: and [Fe/H]$_{CG97}$ for a strong-line star, its metallicity would be

914: underestimated in 0.3 dex. Similar result is obtained for a weak-line

915: star. We conclude therefore, that the effects on the derived

916: metallicity due to a poor fit to the line or the non-linearity of

917: the metallicity scale are comparable.

918: 

919: \subsection{The role of Age in the $W'_V$ ($W'_I$) versus [Fe/H] relationship}

920: 

921: \citet{pont04} investigated the influence of age in the $W'_V$ ($W'_I$)

922: versus [Fe/H] relationship from a theoretical point of view. They used

923: the theoretical calculations of CaT equivalent widths for different

924: values of $\log g$, T$_{eff}$ and metallicity calculated by

925: \citet{jcj92} together with the Padova stellar evolution models

926: \citep{girardi02}. They concluded that the variation of $W'$ with age

927: for a fixed metallicity would be negligible for clusters older than 4

928: Gyr.  However, this was not the case for the younger clusters. This is

929: observed clearly in Figure 15 by \citet{pont04}. For a given

930: metallicity, the sequences in the M$_V$--$\Sigma Ca$ and M$_I$-$\Sigma

931: Ca$ planes are separated as a function of their ages for clusters

932: younger than $\sim$4 Gyr. According to this calculation, for the same

933: metallicity, $W'$ decreases with age. Thus, metallicities for clusters

934: younger than 4 Gyr, calculated from calibrations computed from old

935: stars, will be underestimated. This age dependence is more important

936: in the M$_V$--$\Sigma Ca$ plane than in the M$_I$--$\Sigma Ca$

937: one. This means that $W'_I$ would be less sensitive to age than

938: $W'_V$.

939: 

940: Using the \citet{jcj92} models and the BaSTI stellar evolution models

941: \citep{pie04}, we have estimated the expected $W'$ differences as a

942: function of age. From these calculations, for two clusters with the

943: same metallicity and age 10.5 and 0.6 Gyr respectively, the youngest

944: cluster $W'_V$ would be approximately 0.7 \AA\ lower than that of the

945: oldest one. This implies that the metallicity obtained for young

946: clusters using this calibration would be 0.25 dex more metal-poor than

947: the actual metallicity. In the case of $W'_I$, the difference would be

948: 0.4 \AA, so the metallicity obtained for young clusters would be 0.15

949: dex more metal-poor than the actual one. As we can see in Figure 15 by

950: \citet{pont04}, the difference would be similar for different

951: metallicities.

952:   

953: From our data, we confirm that the influence of age is weak. In Figure

954: \ref{agetest} we plot $W'_I$ versus age for clusters with $-0.17\leq$

955: [Fe/H]$_{CG97}\leq$ +0.07. We have selected this range because it

956: contains clusters with a wide range of ages and is small enough for

957: the metallicity differences to be within the uncertainties. We can see

958: that clusters with ages younger than 5 Gyr (NGC 2141, NGC 2682, NGC

959: 6819 and NGC 7789) have similar $W'_I$ than the oldest one (NGC 6528).

960: There are only two clusters that deviate widely from the behaviour of

961: the others. One of these is the youngest cluster, NGC 6705, which has

962: a larger $W'_I$ than the oldest clusters. This is contrary to the

963: theoretical prediction that it should be smaller. However, we have to

964: take into account that differences of 0.5 \AA\ in $W'_I$ mean

965: differences of $\sim$0.1 dex in [Fe/H]. So the observed variations are

966: similar to the uncertainty in the determination of [Fe/H]. Our data

967: are not accurate enough to detect the influence of age because the

968: uncertainty in the metallicity determination of clusters is similar to

969: the expected variations due to age.

970: 

971: \placefigure{agetest}

972: 

973: \subsection{The influence of [Ca/Fe] abundance}

974: 

975: The CaT has traditionally been used to infer Iron abundances from Ca

976: lines, and we also do so in this paper.  But, the CaT lines strength

977: should also be sensitive to the Ca abundances. In fact, the

978: relationships obtained in this work and those found in the literature

979: have been obtained assuming implicitly the specific relationship

980: between Ca and Fe followed by clusters used in the calibration (see

981: Figure \ref{cafe} for the relationship of the clusters used in this

982: work). Using these relationships to derive Fe abundances in stellar

983: systems with a different chemical evolution than the Milky Way,

984: reflected in the calibrating cluster sample, could give wrong

985: results.

986: 

987: In general, the relationship between the reduced equivalent width of

988: an atomic line and the chemical abundance of the corresponding element

989: is described by the curve of growth. This is only linear for very weak

990: and unsaturated lines. This is not the case for the CaT. As we can

991: find the [Ca/H] ratio for most of the clusters in our sample from the

992: literature, in Figure \ref{cafe} we have plotted $W'_V$ and $W'_I$

993: versus [Ca/H]. The relationship between both is equivalent to the

994: curve of growth. The relations obtained are:

995: 

996: \begin{eqnarray}

997: &[Ca/H]^V&=-2.51(\pm0.08)+0.30(\pm0.01)W'_V~~~\sigma=0.11\label{cahv}\\

998: &[Ca/H]^I&=-2.36(\pm0.08)+0.31(\pm0.01)W'_I~~~\sigma=0.11\label{cahi}

999: \end{eqnarray} 

1000: 

1001: As in the case of the [Fe/H] relationship, we obtain a linear

1002: dependence. However, note that in this case the errors of the fit are

1003: larger. This may be related to the inhomogeneity of the [Ca/H]

1004: abundances, which were obtained from different sources.

1005: 

1006: In any case, even though [Ca/H] changes linearly with $W'$, [Fe/H]

1007: does not have to do likewise. However, as we see in Figures \ref{calv}

1008: and \ref{calI}, the relationship between [Fe/H] and $W'$ is also

1009: linear. On the other hand, since the [Ca/H] and [Ca/Fe] abundances are

1010: related according to [Fe/H] = [Ca/H] $-$ [Ca/Fe], we can expect that

1011: [Ca/Fe] also changes linearly with $W'$ (and with [Fe/H]), if the

1012: relation with [Ca/H] is linear. In fact, in Figure \ref{cafe} we can

1013: check that this is the case over the whole range of [Fe/H] except for

1014: the most metal-poor clusters. Note however that the linear behaviour

1015: of $W'$ with [Ca/H] and [Ca/Fe] is a characteristic of our particular

1016: sample, but this would not have to be the rule.

1017: 

1018: The problem of the relation between the CaT, [Ca/H] and [Fe/H] has

1019: been addressed by \citet{idiart97} from an empirical point of

1020: view. For their sample of late-type stars (G and K), they found that

1021: the dominant stellar parameter controlling the behaviour of the CaT

1022: lines is metallicity, and contrary to what would be expected, the

1023: [Ca/Fe] ratio has practically no effect on the CaT index. However, all

1024: the stars in their sample follow the same relationship between Ca and

1025: Fe, so they cannot check in a general way the influence of the

1026: [Ca/Fe] ratio.

1027: 

1028: To properly investigate the influence of the [Ca/Fe] ratio, it is

1029: necessary to have objects with the same metallicities and different

1030: [Ca/H] ratios. In our sample, most of the metal-poor clusters have

1031: high $\alpha$-element abundances relative to Fe, as is the case for

1032: Ca. On the other hand, open clusters are metal-rich and have low

1033: $\alpha$-element abundances. To study the influence of the [Ca/Fe]

1034: ratio on the CaT calibration as a function of metallicity it would be

1035: necessary to include metal-rich objects with high $\alpha$-element

1036: abundances (i.e.\ stars in the Milky Way bulge) and metal-poor objects

1037: with low $\alpha$-element abundances (i.e.\ perhaps stars in dwarf

1038: galaxies). This sort of work would need a huge observational effort,

1039: which explains why it has not been done until now.

1040: 

1041: \placefigure{cafe}

1042: 

1043: \section{Derived cluster Metallicities\label{derivedmetallicities}}

1044: 

1045: We will use the relationships derived in previous sections to estimate

1046: the metallicities in the three observed clusters without previous

1047: determinations. In fact, we have observed Collinder 110, a poorly

1048: studied cluster with no previous spectroscopic metallicity

1049: determinations. For Berkeley 39, only \citet{f02} have determined its

1050: metallicity from low-resolution spectroscopy. The sequences of these

1051: clusters in the M$_I$--$\Sigma Ca$ plane have been plotted in Figure

1052: \ref{msigmacatest}.

1053: 

1054: \placefigure{msigmacatest}

1055: \subsection{Berkeley 39}

1056: 

1057: The first colour--magnitude diagram of this open cluster was published

1058: by \citet{kaluzny89}. These authors calculated a distance modulus of

1059: (m-M)$_V$ = 13.4 and $E(B-V)=0.12$. These values agree with the

1060: determinations of \citet{carraro94}, who also used colour--magnitude

1061: diagrams. The age of this cluster is 7 $\pm$ 1 Gyr \citep{swp04}.

1062: 

1063: There are few determinations of its metallicity. From photometric data

1064: \citet{twarog97} estimated [Fe/H] = $-0.18\pm 0.03$, while from

1065: low-resolution spectroscopy, \citet{f93} and \citet{f02} obtained

1066: [Fe/H] = $-0.32 \pm 0.08$ and [Fe/H] = $-0.26\pm 0.09$

1067: respectively. In our case we have 10 RGB stars which are cluster

1068: members from their radial velocity, although only 5 stars have $I$

1069: magnitudes available. Moreover, only 2 are brighter than M$_I$ = 0;

1070: nevertheless, the other 3 have magnitudes close to this value. We

1071: therefore used all 5 stars. From Equation \ref{cg97i} we obtain

1072: [Fe/H]$_{CG97}$ = $-0.14 \pm 0.02$. We have used the relationship as a

1073: function of M$_I$ because the RGB is more resolved in the $I$ filter,

1074: and this relation is less sensitive to age.  The calculated value is

1075: slightly more metal-rich than previous spectroscopic

1076: determinations. In the KI03 and ZW84 scales we obtain [Fe/H]$_{KI03}$

1077: = $-0.33 \pm 0.14$ and [Fe/H]$_{ZW84}$ = $-0.23 \pm 0.25$ respectively

1078: from Equations \ref{ki03im} and

1079: \ref{zw84i}. On these scales we have no young and/or metal-rich reference clusters, but, as we have checked before,

1080: the influence of age is weak.

1081: 

1082: We have also calculated the radial velocity of this cluster. We find

1083: V$_r$ = 59 $\pm$5 km s$^{-1}$, which is similar to values found

1084: previously \citep[i.e.][V$_r$=55$\pm$7 Km s$^{-1}$]{f02}.

1085: 

1086: \subsection{Trumpler 5}

1087: 

1088: Trumpler 5, also named Collinder 105, is also a poorly studied

1089: cluster, even though it was discovered about 75 yr ago. It is located

1090: towards the Galactic anticentre in a rich star field in Monoceros, and

1091: in a region of variable interstellar reddening. This has complicated

1092: the studies of this cluster. In fact, only photometric studies could

1093: be found in the literature \citep[e.g.][]{kaluzny98,kimsung03, pca04}

1094: with the exception of the work by \citet{c04}, who observed the CaT

1095: lines in a few stars on the RGB and derived the first spectroscopic

1096: determination of its metallicity. The distance modulus and reddening

1097: of this cluster have been derived from isochrone fitting. Most studies

1098: converge on a reddening of $E(B-V)=0.6$

1099: \citep[e.g.][]{kimsung03}. However, this does not happen in the case

1100: of the distance, where the values lie between (m-M)$_0$ = 12.25

1101: \citep{pca04} and 12.64 \citep{kimsung03}, corresponding to a distance

1102: from the Sun of 2.4 or 3.4 kpc respectively. Also, the age and

1103: metallicity have traditionally been estimated from isochrones. The age

1104: of this cluster is estimated between 2.4 $\pm$ 0.2 \citep{kimsung03}

1105: and 5.0 $\pm$ 05 Gyr \citep{pca04}, while the derived metallicity is

1106: [Fe/H] = $-0.30 \pm 0.15$ dex \citep[e.g.][]{kimsung03,pca04}.

1107: 

1108: We have observed 21 stars in the field of Trumpler 5, 17 of which are

1109: radial velocity members (Table~\ref{starsample}). The metallicity

1110: derived from Equation \ref{cg97i} is [Fe/H]$_{CG97}$ = $-0.36 \pm

1111: 0.05$, which is more metal-rich (although within the error) than the

1112: previous spectroscopic determination, [Fe/H] = $-0.56 \pm 0.11$, by

1113: \citet{c04}. The alternative determination of the metallicity on the

1114: KI03 and ZW84 scales gives [Fe/H]$_{KI03}$ = $-0.56

1115: \pm 0.09$ and [Fe/H]$_{ZW84}$ = $-0.48 \pm 0.20$ respectively from

1116: Equations~\ref{ki03im} and \ref{zw84i}

1117: 

1118: From our data we have also calculated the radial velocity of this

1119: cluster, V$_r$ = 44 $\pm$ 10 km s$^{-1}$, which is similar to the

1120: value derived by \citet[][V$_r$=54$\pm$5 Km s$^{-1}$]{c04}.

1121: 

1122: \subsection{Collinder 110}

1123: 

1124: Collinder 110 is a poorly populated cluster, even less studied than

1125: Trumpler 5. Only two photometric studies can be found in the

1126: literature for the last three decades. Using synthetic

1127: colour--magnitude diagrams, \citet{bragagliatosi03} have estimated a

1128: reddening of 0.38 $\leq E(B-V) \leq 0.45$ and distance modulus

1129: (m-M)$_0$ between 11.8 and 11.9. From these values they derived an age

1130: between 1.1 and 1.5 Gyr. Similar values were found by

1131: \citet{dawson98}. There are no metallicity determinations for this

1132: cluster in the literature. \citet{bragagliatosi03} tried to derive the

1133: metallicity of this cluster from different stellar evolution models,

1134: but concluded that the final result vary widely depending on the

1135: models.

1136: 

1137: The metallicity derived from Equation \ref{cg97i} is [Fe/H]$_{CG97}$ =

1138: $-0.01 \pm 0.07$. If we use Equations \ref{ki03im} and

1139: \ref{zw84i} on KI03 and ZW84 metallicity scales we find [Fe/H]$_{KI03}$ 

1140: = $-0.19 \pm 0.21$ and [Fe/H]$_{ZW84}$ = 0.00 $\pm$ 0.30. From our

1141: data we can also provide the first determination of its radial

1142: velocity, V$_r$ = 45 $\pm$ 8 km sec$^{-1}$.

1143: 

1144: \section{Summary\label{conclusions}}

1145: 

1146: We have observed the CaT lines in RGB stars in a sample of 29 clusters

1147: of the Milky Way. This sample covers an age range of (13 $\leq$

1148: Age/Gyr $\leq$ 0.25) and metallicity range of ($-2.2\leq$ [Fe/H]

1149: $\leq$ +0.47). These are the widest ranges of ages and metallicities

1150: in which the behaviour of the CaT has been investigated in a

1151: homogeneous way until now. We have obtained relationships between the

1152: CaT equivalent widths and metallicities on the scales of \citet{zw84},

1153: \citet{cg97} and \citet{ki03}. The influence of other parameters, such

1154: as age and [Ca/Fe] ratio, has been investigated. Moreover, for the

1155: first time, the behaviour of the CaT lines as a function of luminosity

1156: along the RGB has been studied for the whole range of metallicities in

1157: our sample.

1158: 

1159: The main results of this work are:

1160: \begin{itemize}

1161: 

1162: \item Theoretically, it has been predicted that the sequences of clusters in 

1163: the Luminosity--$\Sigma Ca$ plane may not be linear, and that the

1164: slope should change with metallicity. In this article we have

1165: demonstrated that the nonlinear tendency and the change of the slope

1166: can be (marginally) detected if a wide range of magnitudes in the RGB

1167: is observed.

1168: 

1169: \item However, this behaviour is not significant if only the usual range 

1170: of 3-4 magnitudes below the tip of the RGB is observed. For this

1171: reason, for stars with M$_V\leq$ 1.25 or M$_I\leq$ 0, we have

1172: considered that the sequences of the clusters in the M$_V$--$\Sigma

1173: Ca$ and M$_I$--$\Sigma Ca$ planes are linear, and share a common slope,

1174: independently of metallicity.

1175: 

1176: \item We have obtained relationships between the reduced equivalent 

1177: width ($W'_V$ and $W'_I$) and metallicity on the \citet{zw84},

1178: \citet{cg97} and \citet{ki03} scales. While on the \citet{cg97} and

1179: \citet{ki03} scales these relationships are linear, in the case of the

1180: \citet{zw84} scale, it is quadratic.

1181: 

1182: \item Theory predicts that the relationship between the CaT line equivalent 

1183: widths and metallicity might be dependent on age, mainly for clusters

1184: younger than 4 Gyr. We have studied the influence of age and found

1185: that the expected differences due to age are similar to the

1186: metallicity resolution of our work.

1187: 

1188: \item We have also investigated the influence of Ca abundances on the 

1189: relationships between $W'_V$ and $W'_I$ and metallicity. We have found

1190: that [Ca/H] also changes linearly with $W'_V$ and $W'_I$.

1191: 

1192: \item Finally, the  relationships obtained have been used to compute the 

1193: metallicity of 3 clusters in our sample: Berkeley 39, Trumpler 5 and

1194: Collinder 110. For the last one, there are no previous determinations

1195: of its metallicity in the literature.

1196: 

1197: \end{itemize}

1198: 

1199: \acknowledgments

1200: 

1201: We warmly thank Dr. Antonio Aparicio for many fruitful discussions on

1202: this paper, and a careful and critical reading of the

1203: manuscript. Extensive use was made of the WEBDA database, maintained

1204: at the university of Geneva, Switzerland. C.G. and R.C. acknowledge

1205: the support from the Spanish Ministry of Science and Technology (Plan

1206: Nacional de Investigaci\'on Cient\'{\i}fica, Desarrollo, e

1207: Investigaci\'on Tecnol\'ogica, AYA2004-06343). E. P. acknowledge

1208: support from the Italian MIUR (Ministero dell'Universit\`a e della

1209: Ricerca) under PRIN 2003029437 entitled "Continuities and

1210: discontinuites in the formation of the galaxy". R.Z. acknowledges the

1211: support of the NSF under grant AST05-07364.  

1212: 

1213: Facilities:

1214: \facility{VLT(FORS2)}, \facility{CAHA2.2m(CAFOS)},

1215: \facility{CTIO4m(HYDRA)}, \facility{WHT(WYFFOS)},

1216: \facility{WHT(ISIS)},

1217: \facility{INT(IDS)}.

1218: 

1219: 

1220: %% The reference list follows the main body and any appendices.

1221: %% Use LaTeX's thebibliography environment to mark up your reference list.

1222: %% Note \begin{thebibliography} is followed by an empty set of

1223: %% curly braces.  If you forget this, LaTeX will generate the error

1224: %% "Perhaps a missing \item?".

1225: %%

1226: %% thebibliography produces citations in the text using \bibitem-\cite

1227: %% cross-referencing. Each reference is preceded by a

1228: %% \bibitem command that defines in curly braces the KEY that corresponds

1229: %% to the KEY in the \cite commands (see the first section above).

1230: %% Make sure that you provide a unique KEY for every \bibitem or else the

1231: %% paper will not LaTeX. The square brackets should contain

1232: %% the citation text that LaTeX will insert in

1233: %% place of the \cite commands.

1234: 

1235: %% We have used macros to produce journal name abbreviations.

1236: %% AASTeX provides a number of these for the more frequently-cited journals.

1237: %% See the Author Guide for a list of them.

1238: 

1239: %% Note that the style of the \bibitem labels (in []) is slightly

1240: %% different from previous examples.  The natbib system solves a host

1241: %% of citation expression problems, but it is necessary to clearly

1242: %% delimit the year from the author name used in the citation.

1243: %% See the natbib documentation for more details and options.

1244: 

1245: \begin{thebibliography}{}

1246: \bibitem[Alcaino(1974)]{alcaino74} Alcaino, G., 1974, \aaps, 13, 55

1247: \bibitem[Alcaino \& Liller(1980)]{alcainoliller80} Alcaino, G., \& Liller, W. 1980, \aj, 96, 92

1248: \bibitem[Alcaino \& Liller(1986)]{alcainoliller86} Alcaino, G., \& Liller, W. 1986, \aap, 161, 61

1249: \bibitem[Armandroff \& Zinn(1988)]{az88} Armandroff, T. E., \& Zinn, R. 1988, \aj, 96, 92

1250: \bibitem[Armandroff \& Da Costa(1991)]{adc91} Armandroff, T. E. \& Da Costa, G. S. 1991, \aj, 101, 1329

1251: \bibitem[Bragaglia et al.(2001)]{bra01} Bragaglia, A., Carretta, E., Gratton, R. G., Tosi, M., Bonanno, G., Bruno, P., Cal\'{\i}, A., Claudi,

1252: R., Cosentino, R., Desidera, S., Farisato, G., Rebeschini, M. \& Scuderi, S. 2001, \aj, 121, 327

1253: \bibitem[Bragaglia \& Tosi(2003)]{bragagliatosi03} Bragaglia, A., \& Tosi, M. 2003, \mnras, 343, 306

1254: \bibitem[Brown et al.(1999)]{brown99}Brown, J. A., Wallerstain, G., \& Gonzalez, G. 1999, \aj, 118, 1245

1255: \bibitem[Buonanno et al.(1983)]{buonanno83}Buonanno, R., Buscema, G., Corsi, C. E., Iannicola, M. \& Fusi Pecci, F. 1983, \aaps, 51, 83

1256: \bibitem[Burkhead et al.(1972)]{burkhead72}Burkhead, M. S., Burgess, R. D., \& Haisch, B. M. 1972, \aj, 77, 661

1257: \bibitem[Carraro et al.(1994)]{carraro94} Carraro, G. Chiosi, C., Bressan, A., \& Bertelli, G. 1994, \aaps, 103, 375

1258: \bibitem[Carraro et al.(2001)]{carraro01} Carraro, G. Hassan, S. M., Ortolani, S., \& Vallerini, A. 2001, \aap, 372, 879

1259: \bibitem[Carrera et al.(2007)]{carrera05} Carrera, R., Gallart, C., Hardy, E., Aparicio, A., Zinn, R. 2007, \aj, In preparation

1260: %\bibitem[Carrera et al.(2007b)]{carrera07} Carrera, R.  2006, \aj, In preparation SMC

1261: \bibitem[Carretta \& Gratton(1997)]{cg97} Carretta, E., \& Gratton, R. G. 1997, \aaps, 121, 95 (CG97)

1262: \bibitem[Carretta et al.(2001)]{ccgb01} Carretta, E., Cohen, J. G., Gratton, R. G. \& Behr, B. B. 2001, \aj, 122, 1469

1263: \bibitem[Carretta et al.(2004)]{cbgt04} Carretta, E., Bragaglia, A., Gratton, R. G. \& Tosi, M. 2004, \aap, 422, 951

1264: \bibitem[Cenarro et al.(2001)]{cen01} Cenarro, A. J., Cardiel, N., Gorgas, J., Peletier, R. F., Vazdekis, A., \& Prada, F. 2001, \mnras, 326,

1265: 959

1266: \bibitem[Cole et al.(2000)]{c00} Cole, A. A., Smecker-Hane, T. A., \& Gallagher III, J. S. 2000, \aj, 120, 1808 

1267: \bibitem[Cole et al.(2004)]{c04} Cole, A. A., Smecker-Hane, T. A., Tolstoy, E., Bosler, T. L., \& Gallagher III, J. S. 2004, \mnras, 347, 367

1268: \bibitem[Cole et al.(2005)]{c05}Cole, A. A., Tolstoy, E., Gallagher III, J. S., \& Smecker-Hane, T. A., 2005, \aj, 129, 1465

1269: %\bibitem[Da Costa \& Armandroff(1995)]{dca95} Da Costa, G. S. \& Armandroff, T. E. 1995, \aj, 109, 2533

1270: \bibitem[D\'{\i}az et al.(1989)]{dtt89} D\'{\i}az, A. I., Terlevich, E. \& Terlevich R. 1989, \mnras, 239, 325

1271: \bibitem[Dawson \& Ianna(1998)]{dawson98} Dawson, D. W., \& Ianna, P. A. 1998, \aj, 115, 1076

1272: \bibitem[Erdelyi-Mendes \& Barbuy(1991)]{erdelyi91} Erdelyi-Mendes, M., \& Barbuy, B. 1991, \aap, 241, 176

1273: \bibitem[Feltzing \& Johnson(2002)]{feltzingjohnson02}Feltzing, S., \& Johnson, R. A. 2002, \aap, 385, 67

1274: \bibitem[Friel (1989)]{friel89}Friel, E. D. 1989, \pasp, 101, 244

1275: \bibitem[Friel \& Janes(1993)]{f93}Friel, E. D., \& Janes, K. A. 1993, \aap, 267, 75

1276: \bibitem[Friel et al.(2002)]{f02}Friel, E. D., Janes, K. A., Tavarez, M., Scott, J., Katsanis, R., Lotz, J., Hong, L., \& Miller, N. 2002,

1277: \aj, 124, 2693

1278: %\bibitem[Friel et al.(2005)]{f05} Friel, E. D., Jacobson, H. R. \& Pilachowski, C. A. 2005, \aj, 129, 2725 

1279: \bibitem[Fullton(1996)]{fullton96} Fullton, L. K. 1996, \pasp, 108, 545

1280: \bibitem[Gallart et al.(2001)]{gallart01} Gallart, C., Mart\'{\i}nez-Delgado, D., G\'omez-Flechoso, M. A., \& Mateo, M. 2001, \aj, 121,

1281: 2572

1282: \bibitem[Gim et al.(1998)]{gim98} Gim, M., Vanderverg, D. A., Stetson, P. B., Hesser, J. E., \& Zurek, D. R. 1998, \pasp, 110, 1318

1283: \bibitem[Girardi et al.(2002)]{girardi02}Girardi, L., Bertelli, G., Bressan. A., Chiosi, C., Groenewegen, M: A. T., Marigo, P., Salasnich, B.,

1284: \& Weiss, A. 2002, \aap, 391, 195

1285: \bibitem[Gonzalez \& Wallerstein(2000)]{gonzalezwallerstein00}Gonzalez, G., \& Wallerstein, G. 2000, \pasp, 112, 1081

1286: \bibitem[Gratton(1987)]{gratton87}Gratton, R. G. 1987, \aap, 179, 181

1287: \bibitem[Gratton(2000)]{gratton00} Gratton, R. G. 2000, on ASP Conf. Ser. 198, Stellar Clusters and Associations, ed. R. Pallavicini, G.

1288: Micela \& S. Sciortino (San Francisco: ASP), 225

1289: %\bibitem[Gratton et al.(2004)]{gratton04} Gratton, R. G., Bragaglia, A., Clementini, G., Carretta, E., Di Fabrizio, L., Maio, M., \&

1290: %Taribello, E. 2004, \aap, 421, 937

1291: \bibitem[Gratton \& Contarini(1994)]{gratton94} Gratton, R.~G., 

1292: \& Contarini, G.\ 1994, \aap, 283, 911 

1293: \bibitem[Gratton et al.(2006)]{gratton06} Gratton, R. G., Bragaglia, A., Carretta, E., \& Tosi, M. 2006,

1294: \apj, 642, 462

1295: \bibitem[Gratton \& Ortolami(1989)]{gratton89} Gratton, R. G., \& Ortolami, S. 1989, \aap, 211,41

1296: \apj, 642, 462

1297: %\bibitem[Gratton et al.(2004b)]{gsc04} Gratton, R. G., Sneden, C., \& Carretta, E. 2004b, \araa, 42, 385

1298: \bibitem[Harris \& Zaritsky(2006)]{hz06} Harris, J. \& Zaritsky, D. 2006, \aj, 131, 2514 

1299: \bibitem[Harris(1975a)]{harris75a}Harris, W. E. 1975a, \aj, 82, 954

1300: \bibitem[Harris(1975b)]{harris75b}Harris, W. E. 1975b, \apjs, 29, 397

1301: \bibitem[Harris(1982)]{harris82}Harris, W. E. 1982, \apjs, 50, 573

1302: \bibitem[Harris(1996)]{harris96}Harris, W. E. 1996, \aj, 112, 1487

1303: \bibitem[Hubbs et al.(1990)]{hubbs90}Hubbs, L. M., Thorburn, J. A. \& Rodriguez-Bell, T. 1990, \aj, 100, 710

1304: \bibitem[Idiart et al.(1997)]{idiart97} Idiart, T. P., Th\'evenin, F. \& de Freitas Pacheco, J. A.. 1997, \aj, 113, 1066

1305: \bibitem[Irwin \& Tolstoy(2002)]{irwint02} Irwin, M., \& Tolstoy, E. 2002, \mnras, 336, 643

1306: \bibitem[J\o rgensen et al.(1992)]{jcj92} J$\o$rgersen, U. G., Carlsson, M., \& Johnson, H. R. 1992, \aap, 254, 258

1307: \bibitem[Kaluzny \& Richtler(1989)]{kaluzny89}Kaluzny, J., \& Richtler, T. 1989, \actaa, 39, 139

1308: %\bibitem[Kaluzny \& Mazur(1991)]{kaluzny91}Kaluzny, J., \& Mazur, B. 1991, \actaa, 41, 167

1309: \bibitem[Kaluzny(1998)]{kaluzny98}Kaluzny, J., 1998, Astron. Astrophys. Suppl. Ser. 133, 25

1310: \bibitem[Kassis et al.(1997)]{kassis97}Kassis, M., Janes, K. A., Friel, E. D., \& Phelps, R. L. 1997, \aj, 113, 1723

1311: \bibitem[Kim \& Sung(2003)]{kimsung03}Kim, S. C., \& Sung, H., 2003, J. Korean Astron. Soc., 36, 13

1312: %\bibitem[Kontizas et al.(1993)]{kon93}Kontizas, M., Kontizas, E., \& Michalitsiano, A. G. 1993, \aap, 269, 107

1313: \bibitem[Kraft \& Ivans(2003)]{ki03}Kraft, R. P., \& Ivans, I. I. 2003, \pasp, 115, 143 (KI03)

1314: \bibitem[Layden \& Sarajedini(1997)]{laydensarajedini97}Layden, A. C., \& Sarajedini, A. 1997, \apj, 486, L110

1315: \bibitem[Lee(1977)]{lee77}Lee, S. W. 1977, \aaps, 27, 381 

1316: \bibitem[Lee et al.(1999)]{lee99}Lee, S. K., Kang, Y. W., \& Ann, H. B. 1999, PKAS, 14, 61

1317: \bibitem[Marconi et al.(1997)]{marconi97}Marconi, G., Hamilton, D., Tosi, M., \& Bragaglia, A. 1997, \mnras, 291, 763

1318: \bibitem[Massey et al.(1992)]{massey92}Massey, P., Valdes, F., \& Barnes, J. 1992, A User's Guide to Reducing Slit Spectra with IRAF 

1319: \bibitem[Mathieu(1985)]{mathieu85}Mathieu,  R.  D.  1985, IAUS, 113, 427

1320: \bibitem[Mathieu et al.(1986)]{mathieu86}Mathieu, R. D., Latham, D. W., Griffin, R. F., \& Gunn, J. E. 1986, \aj, 92, 1100

1321: \bibitem[McWilliam et al.(1992)]{mcwilliam92} McWilliam, A., 

1322: Geisler, D., \& Rich, R.~M.\ 1992, \pasp, 104, 1193 

1323: \bibitem[Mermilliod(1995)]{webda}Mermilliod, J. C. 1995 on "Information and On-line Data in Astronomy", Eds. D. Egret \& M. A. Albrecht

1324: (Kluwer Academic Press, Dordrecht) p. 127 (http://obswww.unige.ch/webda)

1325: \bibitem[No\"el et al.(2007)]{noel07}No\"el, N., Gallart, C., Costa, E. \& M\'endez, R.A. 2007, \aj, accepted.

1326: \bibitem[Olszewski et al.(1991)]{ol91} Olszewski, E. W., Schommer, R. A., Suntzeff, N. B., \& Harris, H., C. 1991, \aj, 101, 515

1327: \bibitem[Origlia et al.(2005)]{origlia05} Origlia, L., Valenti, E., \& Rich, R. M. 2005, \aap, 356, 1276

1328: \bibitem[Ortolani et al.(1992)]{ortolani92} Ortolani, S., Bica, E., \& Barbuy, B. 1992, \aaps, 92, 441

1329: \bibitem[Piatti et al.(2004)]{pca04} Piatti, A. E., Calri\'{a}, J. J., \& Ahumada, A. V. 2004, \mnras, 349, 641

1330: \bibitem[Pietrinferni et al.(2004)]{pie04}Pietrinferni, A., Cassisi, S., Salaris, M., \& Castelli, F. 2004, \apj, 612, 168

1331: \bibitem[Pont et al.(2004)]{pont04} Pont, F., Zinn, F., Gallart, C., Hardy, E., \& Winnick, R. 2004, \aj, 127, 840

1332: \bibitem[Richtler \& Sagar(2001)]{richtlersagar01} Richtler, T., \& Sagar, R. 2001, Bull. Astr. Soc. India, 29, 53

1333: \bibitem[Rosenberg et al.(1999)]{rosenberg99} Rosenberg, A., Saviane, I., Piotto, G., \& Aparicio, A. 1999, \aj, 118, 2306

1334: \bibitem[Rosenberg et al.(2000)]{rosenberg00} Rosenberg, A., Piotto, G., Saviane, I., \& Aparicio, A. 2000, \aaps, 144, 5

1335: \bibitem[Rosenberg et al.(2004)]{rosenberg04} Rosenberg, A., Recio-Blanco, A., \& Garc\'{\i}a-Mar\'{\i}n, M. 2004, \apj, 603, 135

1336: \bibitem[Rosvick(1995)]{rosvick95} Rosvick, J. M. 1995, \mnras, 277, 1379

1337: \bibitem[Rosvick \& Vandenverg(1998)]{rosvickvanderverg98} Rosvick, J. M. \& Vandenverg, D. A. 1998, \aj, 115, 1516

1338: \bibitem[Rutledge et al.(1997a)]{r97a} Rutledge, G. A., Hesser, J. E., Stetson, P. B., Mateo, M., Simard, L., Bolte, M., Friel, E. D., \&

1339: Copin, Y. 1997a, \pasp, 109, 883

1340: \bibitem[Rutledge et al.(1997b)]{r97b} Rutledge, G. A., Hesser, J. E., \& Stetson, P. B. 1997b, \pasp, 109, 907

1341: \bibitem[Salaris \& Weiss(2002)]{sw02}Salaris, M., \& Weiss, A. 2002, \aap, 388, 492

1342: \bibitem[Salaris et al.(2004)]{swp04}Salaris, M., Weiss, A., \& Percival, S. M. 2004, \aap, 414, 163

1343: \bibitem[Sarajedini et al.(1999)]{sarajedini99} Sarajedini, A., von Hippel, T., Kozhurina-Platais, V., \& Demarque, P. 1999, \aj, 118, 2294

1344: \bibitem[Sestito et al.(2006)]{sestito06}Sestito. P., Bragaglia, A., Randich, R., Carretta, E., Prisinzano, 

1345: L. \& Tosi, M. 2006, \aap, 456, 121

1346: %\bibitem[Setteducati \& Weaver(1962)]{setteducati62} Setteducati, A. F., \& Weaver, M. F., 1962 on Newly Found Stellar Clusters (Radio Astr.

1347: %Lab. Univ. of California, Berkeley)

1348: \bibitem[Shetrone \& Keane(2000)]{shetrone00} Shetrone, M.~D., \& 

1349: Keane, M.~J.\ 2000, \aj, 119, 840 

1350: \bibitem[Sneden et al.(1997)]{sneden97} Sneden, C., Kraft, 

1351: R.~P., Shetrone, M.~D., Smith, G.~H., Langer, G.~E., \& Prosser, C.~F.\ 

1352: 1997, \aj, 114, 1964

1353: \bibitem[Stetson(1981)]{stetson81} Stetson, P. B. 1981, \aj, 86, 687

1354: \bibitem[Stetson(2000)]{stetson00} Stetson, P. B. 2000, \pasp, 112, 925

1355: \bibitem[Stetson \& Harris(1977)]{stetsonharris77} Stetson, P. B., \& Harris, W. E. 1977, \aj, 82, 954

1356: \bibitem[Stetson et al.(2003)]{stetsonbg03} Stetson, P. B., Bruntt, H., \& Grundahl, F. 2003, \pasp, 115, 413

1357: \bibitem[Sung et al.(1999)]{sung99} Sung, H., Bessel, M. S., Lee, H. W., Kang, Y. H., Lee, S. W. 1999 \mnras, 310, 982

1358: \bibitem[Suntzeff et al.(1992)]{sunt92} Suntzeff, N. B., Schommer, R. A., Olszewski, E. W., \& Walker, A. R. 1992, \aj, 104, 1743

1359: \bibitem[Suntzeff et al.(1993)]{sunt93} Suntzeff, N. B., Mateo, M., Terndrup, D. M., Olszewski, E. W., Geisler, D., \&

1360: Weller, W. 1993, \apj, 418, 208

1361: \bibitem[Tautvaisiene et al.(2005)]{tautvaisiene05}Tautvaisiene, G., Edvardsson, B., Puzeras, E., \& Ilyin, I. 2005, \aap, 431, 933

1362: \bibitem[Tonry \& Davis(1979)]{td79}Tonry, J., \& Davis, M. 1979, \aj, 84, 1511

1363: \bibitem[Twarog et al.(1997)]{twarog97}Twarog, B. A., Ashman, K. M., Anthony-Twarog, B. J. 1997, \aj, 114, 2556

1364: \bibitem[Valdes(1992)]{valdes92} Valdes, F. 1992, Guide to the HYDRA Reduction task DOHYDRA

1365: %\bibitem[Worthey \& Jowett(2003)]{wj03} Worthey, G., \& Jowett, K. J. 2003 \pasp, 115, 96

1366: \bibitem[Yong et al.(2005)]{young05} Yong, D., Carney, B. W., \& Texeira de Almeida, M. L. 2005 \aj, 130, 597

1367: \bibitem[Zinn \& West(1984)]{zw84} Zinn, R., \& West, M. J. 1984 \apjs, 55, 45 (ZW84)

1368: \bibitem[Zoccali et al.(2004)]{zoccali04} Zoccali, M., Barbuy, B., Hill, V., Ortolani, S., Renzini, A., Bica, E., Monany, Y., Pasquini,

1369: L., Minniti, D., \& Rich, R. M. 2004 \aap, 423, 507

1370: \end{thebibliography}

1371: 

1372: \clearpage

1373: 

1374: %% Use the Figure environment and \plotone or \plottwo to include

1375: %% Figures and captions in your electronic submission.

1376: %% To embed the sample graphics in

1377: %% the file, uncomment the \plotone, \plottwo, and

1378: %% \includegraphics commands

1379: %%

1380: %% If you need a layout that cannot be achieved with \plotone or

1381: %% \plottwo, you can invoke the graphicx package directly with the

1382: %% \includegraphics command or use \plotfiddle. For more information,

1383: %% please see the tutorial on "Using Electronic Art with AASTeX" in the

1384: %% documentation section at the AASTeX Web site,

1385: %% http://www.journals.uchicago.edu/AAS/AASTeX.

1386: %%

1387: %% The examples below also include sample markup for submission of

1388: %% supplemental electronic materials. As always, be sure to check

1389: %% the instructions to authors for the journal you are submitting to

1390: %% for specific submissions guidelines as they vary from

1391: %% journal to journal.

1392: 

1393: 

1394: %% This example uses \plotone to include an EPS file scaled to

1395: %% 80% of its natural size with \epsscale. Its caption

1396: %% has been written to indicate that additional Figure parts will be

1397: %% available in the electronic journal.

1398: \clearpage

1399: 

1400: \begin{figure}

1401: \epsscale{1}

1402: \plotone{./f01.eps}

1403: \caption{Comparison between equivalent widths for stars observed with different telescopes. Small differences are within the

1404: uncertainties.\label{telescopes}}

1405: \end{figure}

1406: 

1407: \clearpage

1408: 

1409: \begin{figure}

1410: \epsscale{1}

1411: \plotone{./f02.eps}

1412: \caption{Spectra of four stars in clusters with different metallicities. The metallicity decreases from top to bottom. Note

1413: how the strength of the Ca {\footnotesize II} triplet lines increases with metallicity.\label{spectra}}

1414: \end{figure}

1415: 

1416: \clearpage

1417: 

1418: \begin{figure}

1419: \epsscale{1}

1420: \plotone{./f03.eps}

1421: \caption{Continuum (clear) and line (dark) bandpasses defined by (a) 

1422: \citet{cen01}, (b) \citet{r97a} and (c) \citet{az88}. They

1423: have been overplotted on to metal-poor (left) and metal-rich (right)

1424: stars. The bands of \citet{cen01} are wider in the lines to cover the

1425: wings fully and narrower in the continuum in order to avoid the most

1426: prominent molecular features for metal-rich stars.\label{bandas}}

1427: \end{figure}

1428: \clearpage

1429: 

1430: \begin{figure}

1431: \epsscale{0.6}

1432: \plotone{./f04.eps}

1433: \caption{Comparison between $\Sigma Ca$, as defined by \citet{adc91}, 

1434: \citet{r97b} and \citet{c04}, and the values obtained in this paper.

1435:  The dashed lines represent the one-to-one equivalence. Solid lines

1436:  are best fits to the data.\label{comp}}

1437: \end{figure}

1438: 

1439: \clearpage

1440: 

1441: \begin{figure}

1442: \epsscale{1}

1443: \plotone{./f05.eps}

1444: \caption{Stars in the M$_V$--$\Sigma Ca$ and M$_I$--$\Sigma Ca$ planes for

1445: the clusters in which we have observed stars along the RGB: NGC 7078

1446: (open circles), NGC 288 (hexagons), NGC 104 (triangles), NGC 2141

1447: (crosses) and NGC 6791 (filled circles). The individual quadratic fit

1448: to each cluster is plotted (solid lines). Dotted lines represent the

1449: extrapolation of the fit in the magnitude range where there are no

1450: calibration stars. We also plotted the theoretical predictions for

1451: each of them (dashed lines). The models have been shifted to match

1452: approximately the cluster sequences (see text for details). Errorbars

1453: are omitted for clarity, but the typical error is shown on the lower

1454: rigth corner.\label{luminosityfig}}

1455: \end{figure}

1456: 

1457: \clearpage

1458: 

1459: \begin{figure}

1460: \epsscale{1}

1461: \plotone{./f06.eps}

1462: \caption{Different fits to the sequences of the clusters in which we have 

1463: observed stars along the RGB in the M$_V$--$\Sigma Ca$ and

1464: M$_I$--$\Sigma Ca$ planes. Solid lines are the quadratic fit to each

1465: cluster independently. Dashed lines are the quadratic fit when the

1466: linear and quadratic terms are the same for all clusters. Finally,

1467: dotted--dashed lines are the linear fits for stars brighter than

1468: M$_V\leq$ 1.25 and M$_I\leq$ 0, assuming the same slope for all

1469: clusters. Dotted lines are the regions in which the fits are

1470: extrapolated. \label{luminositymodels}}

1471: \end{figure}

1472: 

1473: \clearpage

1474: 

1475: \begin{figure}

1476: \epsscale{1}

1477: \plottwo{./f07a.eps}{./f07b.eps}

1478: \caption{Values of the slopes obtained from the individual fit for each 

1479: cluster, versus $W'$. Solid lines are the linear fit, which is given

1480: at the bottom. Note that there is no correlation between slope and

1481: $W'$ (and therefore [Fe/H]) in any of the filters.\label{pendientefig}}

1482: \end{figure}

1483: 

1484: \clearpage

1485: 

1486: \begin{figure}

1487: \epsscale{1}

1488: \plotone{./f08.eps}

1489: \caption{Cluster sample in the M$_V$--$\Sigma Ca$ plane. Solid lines are 

1490: the linear fit to the stars in each cluster when we assume that the

1491: slope is the same for all of them. The typical error is shown on the

1492: lower rigth corner.\label{mvsigmaca}}

1493: \end{figure}

1494: 

1495: \clearpage

1496: 

1497: \begin{figure}

1498: \epsscale{1}

1499: \plotone{./f09.eps}

1500: \caption{Same as Figure \ref{mvsigmaca} but in the M$_I$--$\Sigma Ca$ plane.\label{misigmaca}}

1501: \end{figure}

1502: 

1503: \clearpage

1504: 

1505: \begin{figure}

1506: \epsscale{1}

1507: \plotone{./f10.eps}

1508: \caption{Top panel: [Fe/H] versus $W'_V$. The solid lines are the best 

1509: linear fit to the data. Dashed lines define the confidence band of the

1510: fit. Open circles are clusters younger than 4 Gyr. The residuals of

1511: the linear fit are shown in the bottom panel. Note that the $W'_V$

1512: errors are smaller that the size of points in most cases. The clusters excluded

1513: from the analysis (NGC 2420, NGC 2506 and Berkeley 32) have not been plotted.\label{calv}}

1514: \end{figure}

1515: 

1516: \clearpage

1517: 

1518: \begin{figure}

1519: \epsscale{1}

1520: \plotone{./f11.eps}

1521: \caption{Same as Figure \ref{calv} but with  $W'_I$. \label{calI}}

1522: \end{figure}

1523: 

1524: \clearpage

1525: 

1526: \begin{figure}

1527: \epsscale{1}

1528: \plotone{./f12.eps}

1529: \caption{$W'_V$ (left) and $W'_I$ (right) versus [Fe/H] on the KI03 (top) and the ZW84 (bottom) metallicity scales. The lines are the best

1530: fit to the data. In the case of the ZW84 metallicity scale, a second-order polynomial results in an improvement

1531: of the fit.\label{metallicities}}

1532: \end{figure}

1533: 

1534: \clearpage

1535: 

1536: \begin{figure}

1537: \epsscale{1}

1538: \plotone{./f13.eps}

1539: \caption{$W'_I$ versus age for clusters with $ -0.17\leq$ [Fe/H] $\leq$ 

1540: +0.07. Independently of their ages, all clusters have similar $W'_I$,

1541: with the exception of the youngest cluster, NGC 6705 (0.25 Gyr). \label{agetest}}

1542: \end{figure}

1543: 

1544: \clearpage

1545: 

1546: \begin{figure}

1547: \epsscale{1}

1548: \plotone{./f14.eps}

1549: \caption{[Ca/Fe] versus $W'_V$ (left) and $W'_I$ (right) for the clusters 

1550: in our sample with Ca abundances available.\label{cafe}}

1551: \end{figure}

1552: 

1553: \clearpage

1554: 

1555: \begin{figure}

1556: \epsscale{1}

1557: \plotone{./f15.eps}

1558: \caption{$W'_V$ (left) and $W'_I$ (right) versus [Ca/H] ratio. The solid 

1559: line is the best linear fit to the data.  As before, open circles are

1560: clusters younger than 4 Gyr. The residuals of the linear fit are shown

1561: in the bottom panel.\label{cah}}

1562: \end{figure}

1563: \clearpage

1564: 

1565: \begin{figure}

1566: \epsscale{1}

1567: \plotone{./f16.eps}

1568: \caption{Sequences of clusters Berkeley 39 (hexagons), Trumpler 5 (open 

1569: circles) and Collinder 110 (filled squares) in the M$_I$--$\Sigma Ca$

1570: plane. Solid lines are the linear sequence fits to the data for each

1571: cluster when the same slope is assumed for all clusters. The typical

1572: error is shown on the lower right corner.\label{msigmacatest}}

1573: \end{figure}

1574: 

1575: %% Here we use \plottwo to present two versions of the same Figure,

1576: %% one in black and white for print the other in RGB color

1577: %% for online presentation. Note that the caption indicates

1578: %% that a color version of the Figure will be available online.

1579: %%

1580: 

1581: %\begin{figure}

1582: %%\plottwo{f2.eps}{f2_color.eps}

1583: %\caption{A panel taken from Figure 2 of \citet{rudnick03}. See the electronic edition of the Journal for a color version of this Figure.\label{fig2}}

1584: %\end{figure}

1585: 

1586: %% If you are not including electonic art with your submission, you may

1587: %% mark up your captions using the \figcaption command. See the

1588: %% User Guide for details.

1589: %%

1590: %% No more than seven \figcaption commands are allowed per page,

1591: %% so if you have more than seven captions, insert a \clearpage

1592: %% after every seventh one.

1593: 

1594: %% Tables should be submitted one per page, so put a \clearpage before

1595: %% each one.

1596: 

1597: %% Two options are available to the author for producing tables:  the

1598: %% deluxetable environment provided by the AASTeX package or the LaTeX

1599: %% table environment.  Use of deluxetable is preferred.

1600: %%

1601: 

1602: %% Three table samples follow, two marked up in the deluxetable environment,

1603: %% one marked up as a LaTeX table.

1604: 

1605: %% In this first example, note that the \tabletypesize{}

1606: %% command has been used to reduce the font size of the table.

1607: %% We also use the \rotate command to rotate the table to

1608: %% landscape orientation since it is very wide even at the

1609: %% reduced font size.

1610: %%

1611: %% Note also that the \label command needs to be placed{shetrone00}

1612: %% inside the \tablecaption.

1613: 

1614: %% This table also includes a table comment indicating that the full

1615: %% version will be available in machine-readable format in the electronic

1616: %% edition.

1617: %%

1618: \clearpage

1619: 

1620: \begin{deluxetable}{lcccccccccccc}

1621: \tabletypesize{\scriptsize}

1622: \rotate

1623: \tablewidth{0pt}

1624: \tablecaption{Cluster sample.

1625: \label{clustersample}}

1626: \tablehead{

1627: \colhead{Cluster} & \colhead{[Fe/H]$_{CG97}$} & \colhead{Ref.}& \colhead{[Fe/H]$_{ZW84}$}&

1628: \colhead{[Fe/H]$_{KI03}$} & \colhead{[Ca/H]} & \colhead{Ref.} &

1629: \colhead{Age(Gyr)} & \colhead{Ref.} &\colhead{$(m-M)_V$} &

1630: \colhead{$E(B-V)$} & \colhead{Ref.} & \colhead{Run}}

1631: \startdata{shetrone00}

1632: NGC 104 (47 Tuc) & $-0.67\pm0.03$ & 31 & $-0.71$ & $-0.70$ & $0.47\pm0.02$ & 31 & $10.7\pm1.0$ & 8 & 13.32 & 0.05 & 13 & 7 \\

1633: NGC 188 & $-0.07\pm0.04$ & 2 & \nodata & \nodata & \nodata & \nodata & $6.30\pm0.3$ & 9 & 11.44 & 0.09 & 17 & 3 \\

1634: NGC 288 & $-1.07\pm0.03$ & 1 & $-1.40$ & $-1.41$ & $-0.79\pm0.02$ & 33 & $11.3\pm1.1$ & 8 & 14.64 & 0.03 & 13 & 7 \\

1635: NGC 362 & $-1.09\pm0.03$ & 1 & $-1.27$ & $-1.34$ & $-0.91\pm0.01$ & 33 & $8.7\pm1.5$ & 8 & 14.75 & 0.05 & 13 & 7 \\

1636: NGC 1851 & \nodata & \nodata & $-1.36$ & \nodata & \nodata & \nodata & $9.2\pm1.5$ & 8 & 15.49 & 0.02 & 13 & 7 \\

1637: NGC 1904 (M79) & $-1.37\pm0.05$ & 1 & $-1.69$ & $-1.64$ & $-1.07\pm0.02$ & 32 & $11.7\pm1.3$ & 8 & 15.53 & 0.01 & 13 & 7 \\

1638: Berkeley 20 & $-0.49\pm0.05$ & 22 & \nodata & \nodata & $-0.42\pm0.05$ & 22 & $4.05\pm0.7$ & 9 & 15.84 & 0.38 &  22 & 7 \\

1639: NGC 2141 & $-0.14\pm0.05$ & 22 & \nodata & \nodata & $-0.04\pm0.04$ & 22 & $2.45\pm0.9$ & 9 & 14.15 & 0.40 & 14 & 3,7 \\

1640: Collinder 110 & \nodata & \nodata & \nodata & \nodata  & \nodata & \nodata & $1.3\pm0.2$ & 24 & 13.04 & 0.40 & 24 & 3 \\

1641: Trumpler 5 & \nodata & \nodata & \nodata & \nodata & \nodata & \nodata & $5.7\pm2.3$ & 9 & 14.50 & 0.60 & 23 & 3,7 \\

1642: NGC 2298 & $-1.74\pm0.04$ & 1 & $-1.85$ & $-1.64$ & $-1.35\pm0.03$ & 35 & $12.6\pm1.4$ & 8 & 15.54 & 0.13 & 13 & 7 \\

1643: Berkeley 32 & $-0.29\pm0.04$ & 32 & \nodata & \nodata & \nodata & \nodata & $5.9\pm1.6$ & 9 & 12.85 & 0.08 & 25 & 3 \\

1644: Melote 66 & $-0.38\pm0.06$ & 3 & \nodata & \nodata & $-0.29\pm0.10$ & 40 & $5.3\pm1.4$ & 9 & 13.63 & 0.14 & 26 & 7 \\

1645: Berkeley 39 & \nodata & \nodata & \nodata & \nodata & \nodata & \nodata & $7.0\pm1.0$ & 9 & 13.24 & 0.11 & 26 & 7 \\

1646: NGC 2420 & $-0.44\pm0.15$ & 3 & \nodata & \nodata & \nodata & \nodata & $2.2\pm0.3$ & 9 & 12.0 & 0.05 & 15 & 3 \\

1647: NGC 2506 & $-0.20\pm0.01$ & 31 & \nodata & \nodata & \nodata & \nodata & $2.1\pm0.3$ & 9 & 12.60 & 0.09 & 16 & 3,7 \\

1648: NGC 2682 (M 67) & $-0.03\pm0.03$ & 4 & \nodata & \nodata & $+0.01\pm0.06$ & 4 & $4.3\pm0.5$ & 9 & 9.65 & 0.04 & 17 & 3,7 \\

1649: NGC 3201 & $-1.24\pm0.12$ & 1 & $-1.61$ & $-1.48$ & $-1.11\pm0.02$ & 34 & $11.3\pm1.1$ & 8 & 14.17 & 0.21 & 13 & 4 \\

1650: NGC 4590 (M 68) & $-2.00\pm0.03$ & 1 & $-2.09$ & $-2.43$ & $-1.68\pm0.02$ & 34 & $11.2\pm0.9$ & 8 & 15.14 & 0.04 & 13 & 3,4,6,7 \\

1651: NGC 5927 & \nodata & \nodata & $-0.30$ & \nodata & \nodata & \nodata & $10.9\pm2.2$ & 30 & 15.81 & 0.47 & 13 & 6 \\

1652: NGC 6352 & $-0.64\pm0.02$ & 1 & $-0.51$ & \nodata & $-0.63\pm0.14$ & 36 & $9.9\pm1.4$ & 8 & 14.39 & 0.21 & 13 & 6 \\

1653: NGC 6528 & $-0.17\pm0.02$ & 28,29 & $+0.12$ & \nodata & $-0.03\pm0.01$ & 37 & $11.2\pm2.0$ & 10 & 16.16 & 0.55 & 18 & 7 \\

1654: NGC 6681 (M 70) & $-1.35\pm0.03$ & 1 & $-1.51$ & \nodata & \nodata & \nodata & $11.5\pm1.4$ & 8 & 14.93 & 0.07 & 13 & 7 \\

1655: NGC 6705 (M 11) & $+0.07\pm0.05$ & 5 & \nodata & \nodata & $-0.12\pm0.06$ & 5 & $0.25\pm0.1$ & 11 & 12.88 & 0.43 & 11 & 2,5 \\

1656: NGC 6715 (M 54) & \nodata & \nodata & $-1.42$ & $-1.47$ & $-0.98\pm0.04$ & 39 & $12\pm1.5$ & 27 & 17.77 &  0.16 & 19 & 7 \\

1657: NGC 6791 & $+0.47\pm0.04$ & 6 & \nodata & \nodata & \nodata & \nodata & $12.0\pm1.0$ & 12 & 13.07 & 0.09 & 12 & 5 \\

1658: NGC 6819 & $+0.07\pm0.03$ & 7 & \nodata & \nodata & $+0.03\pm0.06$ & 7 & $2.9\pm0.7$ & 9 & 12.35 & 0.14 & 20,7 & 1,5 \\

1659: NGC 7078 (M 15) & $-2.12\pm0.04$ & 1 & $-2.15$ & $-2.42$ & $-1.88\pm0.02$ & 38 & $11.7\pm0.8$ & 8 & 15.31 & 0.09 & 13 & 7 \\

1660: NGC 7789 & $-0.04\pm0.05$ & 4 & \nodata & \nodata & $+0.10\pm0.08$ & 4 & $1.3\pm0.3$ & 9 & 12.20 & 0.28 & 4,21 & 1,3 \\

1661: \enddata

1662: \tablerefs{

1663: (1) \citet{cg97}; (2) \citet{hubbs90}; (3) \citet{gratton00}

1664: (4) \citet{tautvaisiene05}; (5) \citet{gonzalezwallerstein00}; (6) \citet{gratton06};

1665: (7) \citet{bra01}; (8) \citet{sw02}; (9) \citet{swp04};

1666: (10) \citet{feltzingjohnson02}; (11) \citet{sung99}; (12) \citet{stetsonbg03};

1667: (13) \citet{rosenberg99}; (14) \citet{carraro01}; (15) \citet{lee99};

1668: (16) \citet{marconi97}; (17) \citet{sarajedini99}; (18) \citet{ortolani92};

1669: (19) \citet{rosenberg04}; (20) \citet{rosvickvanderverg98}; (21) \citet{gim98};

1670: (22) \citet{young05}; (23) \citet{kimsung03}; (24) \citet{bragagliatosi03}; 

1671: (25) \citet{richtlersagar01}; (26) \citet{kassis97}; (27) \citet{laydensarajedini97}

1672: (28) \citet{zoccali04}; (29) \citet{origlia05}; (30) \citet{fullton96}; (31) \citet{cbgt04}; (32)

1673: \citet{sestito06}; (33) \citet{shetrone00}; (34) \citet{gratton89}; (35) \citet{mcwilliam92}; (36) \citet{gratton87}; (37) \citet{ccgb01};

1674: (38) \citet{sneden97}; (39) \citet{brown99}; (40) \citet{gratton94}.}

1675: 

1676: \end{deluxetable}

1677: \clearpage

1678: 

1679: %% Any table notes must follow the \end{tabular} command.

1680: 

1681: 

1682: %% If the table is more than one page long, the width of the table can vary

1683: %% from page to page when the default \tablewidth is used, as below.  The

1684: %% individual table widths for each page will be written to the log file; a

1685: %% maximum tablewidth for the table can be computed from these values.

1686: %% The \tablewidth argument can then be reset and the file reprocessed, so

1687: %% that the table is of uniform width throughout. Try getting the widths

1688: %% from the log file and changing the \tablewidth parameter to see how

1689: %% adjusting this value affects table formatting.

1690: 

1691: %% The \dataset macro has also been applied to a few of the objects to

1692: %% show how many observations can be tagged in a table.

1693: 

1694: \begin{deluxetable}{cccccc}

1695: \tablecaption{Observing runs

1696: \label{runs}}

1697: \tablewidth{0pt}

1698: \tablehead{

1699: \colhead{Run} & \colhead{Date} & \colhead{Telescope} & \colhead{Instrument} & \colhead{Resolution}&

1700: \colhead{\AA/pix}}

1701: \startdata

1702: 1 & May 2002 & WHT & ISIS & 7000 & 0.41 \\

1703: 2 & April 2002 & WHT & WYFFOS & 4000 & 1.5 \\

1704: 3 & December 2002 & INT & IDS & 6000 & 0.45 \\

1705: 4 & January 2005 & CTIO 4m & HYDRA & 6000 & 0.9 \\

1706: 5 & June 2005 & CAHA 2.2m & CAFOS & 2000 &2.0 \\

1707: 6 & 2005 & VLT & FORS2 MXU & 5000 & 0.85 \\

1708: 7 & ESO Archive & VLT & FORS2 MXU/MOS & 5000 & 0.85 \\

1709: \enddata

1710: \end{deluxetable}

1711: 

1712: \clearpage

1713: \begin{deluxetable}{lcccccccc}

1714: \tablewidth{0pt}

1715: \tablecaption{Star sample.

1716: \label{starsample}}

1717: \tablehead{

1718: \colhead{Cluster} & \colhead{Id} & \colhead{$\Sigma$Ca} &

1719: \colhead{$\sigma_{\Sigma Ca}$} & \colhead{V} &\colhead{I} &

1720: \colhead{V$_r$} & \colhead{$\sigma_{V_r}$} & \colhead{Comments}}

1721: \startdata

1722: ngc104 & S2701 & 6.18 & 0.13 & 14.07 & 99.99 &  2.22 & 4.90 &  Member?\\

1723: ngc104 & S2703 & 6.55 & 0.07 & 12.99 & 99.99 &  4.44 & 4.34 &  No member\\

1724: ngc104 & S2705 & 7.27 & 0.02 & 12.08 & 99.99 &-13.24 & 3.68 &  \\

1725: ngc104 & S2707 & 6.96 & 0.07 & 13.35 & 99.99 & -2.71 & 5.46 & \\

1726: ngc104 & S2712 & 7.34 & 0.05 & 12.89 & 99.99 &  9.46 & 5.25 &  No member\\

1727: \enddata

1728: \tablerefs{{\bf NGC 104} \citet{lee77} [L], \citet{stetson00} [S];

1729: {\bf NGC 188} Webda http://obswww.unige.ch/webda/ \citep{webda} [W]; {\bf NGC 288} \citet{alcainoliller80} [A];

1730: {\bf NGC 362} \citet{harris82} [H]; {\bf NGC 1851} \citet{stetson81} [S]; {\bf NGC 1904} \citet{stetsonharris77} [S];

1731: {\bf Berkeley 20} Webda [W]; {\bf NGC 2141} \citet{burkhead72} [B]; \citet{rosvick95} [R]; Webda [W];

1732: {\bf Collinder 110} Webda [W]; {\bf Trumpler 5} Webda [W]; {\bf NGC 2298} \citet{alcainoliller86} [A], \citet{alcaino74} [A];

1733: {\bf Berkeley 32} Webda [W]; {\bf Melotte 66} Webda [W]; {\bf Berkeley 39} Webda [W]; {\bf NGC 2420} Webda [W]; 

1734: {\bf NGC 2506} Webda [W]; {\bf NGC 2682} Webda [W]; {\bf NGC 3201} \citet{stetson00} [S]; 

1735: {\bf NGC 4590} \citet{harris75a} [H], \citet{stetson00} [S]; {\bf NGC 5927} Zoccali (private communication); {\bf NGC 6352} Zoccali (private communication);

1736: {\bf NGC 6528} \citet{ortolani92} [O]; {\bf NGC 6681} \citet{harris75b} [H] \citet{rosenberg00} [R];

1737: {\bf NGC 6705} \citet{sung99} [S]; {\bf NGC 6715} \citet{rosenberg04} [R]; {\bf NGC 6791} \citet{stetsonbg03} [S];

1738: {\bf NGC 6819} Webda [W]; {\bf NGC 7078} \citet{buonanno83} [B]; \citet{stetson00} [S]; {\bf NGC 7789} Webda [W];}

1739: \tablecomments{Table \ref{starsample} is published in its enterety in the electronic edition of Astronomical Journal. A portion is shown here for

1740: guidance regarding its form and content.}

1741: 

1742: \end{deluxetable}

1743: \clearpage

1744: 

1745: \begin{deluxetable}{lccccc}

1746: \tabletypesize{\footnotesize}

1747: \tablecaption{Radial velocities of the cluster sample.

1748: \label{samplevr}}

1749: \tablewidth{0pt}

1750: \tablehead{

1751: \colhead{Cluster} & \colhead{V$_r$} & \colhead{$\sigma(V_r)$} &

1752: \colhead{Stars} & \colhead{V$_r$}(ref.) & \colhead{Ref.}}

1753: \startdata

1754: NGC 104 (47 Tuc) & -16 & 11 & 32 & -18.7 & 1 \\

1755: NGC 188 & -44 & 20 &8 & -45 & 2 \\ 	       

1756: NGC 288 & -50 & 11 & 19 & -46.6 & 1 \\	       

1757: NGC 362 & 213 & 7 & 16 & 223.5 & 1 \\	       

1758: NGC 1851 & 321 & 9 & 14 & 320.5 & 1 \\	       

1759: NGC 1904 (M79) & 227 & 5 & 16 & 206 & 1 \\

1760: Berkeley 20 & 80 & 7 & 4 & 70 & 2 \\	       

1761: NGC 2141 & 44 & 10 & 21 & 33/64 & 3,4 \\

1762: Collinder 110 & 45 & 11 & 8 & \nodata & \nodata \\

1763: Trumpler 5 & 44 & 10 & 15 & 54 & 4 \\	   

1764: NGC 2298 & 153 & 15 & 6 & 148.9 & 1 \\	      

1765: Berkeley 32 & 98 & 12 & 3 & 101 & 2 \\	      

1766: Melote 66 & 18 & 10 & 11 & 23 & 5 \\ 	      

1767: Berkeley 39 & 59 & 6 & 5 & 55 & 2 \\	      

1768: NGC 2420 & 69 & 5 & 5 & 67 & 2 \\	

1769: NGC 2506 & 76 & 5 & 3 & 84 & 6 \\	

1770: NGC 2682 (M 67) & 36 & 6 & 9 & 33 & 2 \\

1771: NGC 3201 & 491 & 3 & 10 & 494 & 1 \\ 	      

1772: NGC 4590 (M 68) & -89 & 7 & 19 & -93.4 & 1 \\ 

1773: NGC 5927 & -84 & 5 & 20 & -107.5 & 1 \\	      

1774: NGC 6352 & -114 & 8 & 23 & -121 & 1 \\	      

1775: NGC 6528 & 220 & 7 & 5 & 206 & 1 \\ 	      

1776: NGC 6681 (M 70) & 199 & 7 & 4 & 220 & 1 \\

1777: NGC 6705 (M 11) & 28 & 7 & 10 & 34 & 7 \\

1778: NGC 6715 (M 54) & 156 & 8 & 23 & 142 & 1 \\

1779: NGC 6791 & -46 & 10 & 10 & -57 & 2 \\	      

1780: NGC 6819 & 2 & 5 & 7 & -5 & 2 \\	

1781: NGC 7078 (M 15) & -108 & 10 & 33 & -107 & 1 \\

1782: NGC 7789 & -58 & 6 & 20 & -64 & 2 \\ 	      

1783: \enddata

1784: \tablerefs{

1785: (1) \citet{harris96}; (2) \citet{f02}; (3) \citet{friel89}; (4) \citet{c04} (5) \citet{f93}; (6) \citet{mathieu85};

1786: (7) \citet{mathieu86}}

1787: \end{deluxetable}

1788: 

1789: \clearpage

1790: 

1791: \begin{deluxetable}{cc}

1792: \tablecaption{Line and continuum bandpasses

1793: \label{bandastable}}

1794: \tablewidth{0pt}

1795: \tablehead{

1796: \colhead{Line Bandpasses (\AA)} & \colhead{Continuum bandpasses (\AA)}}

1797: \startdata

1798: 8484-8513 & 8474-8484\\

1799: 8522-8562 & 8563-8577\\

1800: 8642-8682 & 8619-8642\\

1801: \nodata & 8799-8725\\

1802: \nodata & 8776-8792\\

1803: \enddata

1804: \end{deluxetable}

1805: 

1806: \clearpage

1807: 

1808: \begin{deluxetable}{ccccc}

1809: \tabletypesize{\footnotesize}

1810: \tablewidth{0pt}

1811: \tablecaption{Coefficients of the quadratic fits in the form $\Sigma Ca$=$W'$+$\beta$M$_{V,I}$+$\gamma$M$_{V,I}^2$ to the

1812: sequence of each cluster individually. The clusters are ordered by metallicity.\label{luminositytable}}

1813: \tablehead{

1814: \colhead{Cluster} & \colhead{W'} &\colhead{$\beta$} & \colhead{$\gamma$} & \colhead{$\sigma$}}

1815: \startdata

1816: \multicolumn{5}{c}{M$_V$}\\

1817: NGC 6791 & 10.26$\pm$0.97 & -1.64$\pm$1.28 & 0.38$\pm$0.35 & 0.72\\

1818: NGC 2141 & 8.30$\pm$0.14 & -0.91$\pm$0.13 & 0.02$\pm$0.15 & 0.44\\

1819: NGC 104  & 6.77$\pm$0.06 & -0.69$\pm$0.05 & 0.15$\pm$0.04 & 0.30\\

1820: NGC 288  & 5.46$\pm$0.08 & -0.56$\pm$0.05 & 0.07$\pm$0.03 & 0.29\\

1821: NGC 7078 & 2.79$\pm$0.06 & -0.39$\pm$0.04 & 0.15$\pm$0.04 & 0.22\\

1822: \multicolumn{5}{c}{M$_I$}\\

1823: NGC 6791 & 8.49$\pm$0.24 & -0.63$\pm$0.74 & 0.22$\pm$0.79 & 0.54\\

1824: NGC 2141 & 7.52$\pm$0.11 & -0.73$\pm$0.20 & 0.00$\pm$0.11 & 0.47\\

1825: NGC 104  & 6.30$\pm$0.04 & -0.42$\pm$0.05 & 0.06$\pm$0.02 & 0.14\\

1826: NGC 288  & 5.09$\pm$0.07 & -0.44$\pm$0.04 & 0.03$\pm$0.03 & 0.27\\

1827: NGC 7078 & 2.64$\pm$0.08 & -0.27$\pm$0.14 & 0.05$\pm$0.07 & 0.22\\

1828: \tableline

1829: \enddata

1830: %\label{luminositytable}

1831: \end{deluxetable}

1832: 

1833: \clearpage

1834: 

1835: \begin{deluxetable}{lcccc}

1836: \tabletypesize{\footnotesize}

1837: \tablewidth{0pt}

1838: \tablecaption{Derived $W'_V$ and $W'_I$ and number of stars used.\label{fitstable}}

1839: \tablehead{

1840: \colhead{Cluster} & \colhead{$W'_V$} & \colhead{\#Star} & \colhead{$W'_I$} &

1841: \colhead{\#Star}}

1842: \startdata

1843: NGC 104 & $6.94\pm0.01 $ & 34 & $6.23\pm0.01$ & 14\\

1844: NGC 188 & $8.17\pm0.07 $ & 6 & $7.27\pm0.08$ & 5\\

1845: NGC 288 & $5.51\pm0.01 $ & 19 & $5.04\pm0.03$ & 14\\

1846: NGC 362 & $6.01\pm0.01 $ & 16 & \nodata & \nodata\\

1847: NGC 1851 & $5.94\pm0.03 $ & 14 & $5.24\pm0.04$ & 8\\

1848: NGC 1904 & $4.91\pm0.03 $ & 16 & \nodata & \nodata\\

1849: Berkeley 20 & $6.86\pm0.03 $ & 4 & $6.39\pm0.03$ & 3\\

1850: NGC 2141 & $8.33\pm0.01 $ & 18 & $7.67\pm0.02$ & 15\\

1851: Collinder 110 & $8.21\pm0.04 $ & 11 & $7.74\pm0.06$ & 6\\

1852: Trumpler 5 & $7.52\pm0.04 $ & 16 & $6.97\pm0.04$ & 15\\

1853: NGC 2298 & $3.75\pm0.03 $ & 6 & $3.09\pm0.03$ & 5\\

1854: Berkeley 32 & $5.90\pm0.08 $ & 4 & $5.27\pm0.08$ & 4\\

1855: Melote 66 & $7.69\pm0.03 $ & 11 & $6.90\pm0.03$ & 11\\

1856: Berkeley 39 &$8.21\pm0.04 $ & 5 & $7.27\pm0.06$ & 3\\

1857: NGC 2420 & $ 6.26\pm0.09 $ & 6 & $6.15\pm0.08$ & 4\\

1858: NGC 2506 & $ 6.96\pm0.09 $ & 4 & $6.37\pm0.09$ & 3\\

1859: NGC 2682 & $ 8.24\pm0.01 $ & 6 & $7.48\pm0.01$ & 8\\

1860: NGC 3201 & $ 5.46\pm0.03 $ & 9 & $4.76\pm0.02$ & 6 \\

1861: NGC 4590 & $ 2.84\pm0.02 $ & 19 & $2.19\pm0.06$ & 12\\

1862: NGC 5927 & $ 7.81\pm0.01 $ & 21 & $6.92\pm0.01$ & 13\\

1863: NGC 6352 & $ 7.10\pm0.01 $ & 19 & $6.31\pm0.01$ & 19\\

1864: NGC 6528 & $ 8.28\pm0.04 $ & 5 & $7.58\pm0.04$ & 5\\

1865: NGC 6681 & $ 5.05\pm0.03 $ & 4 & $4.49\pm0.07$ & 3\\

1866: NGC 6705 & $ 8.95\pm0.07 $ & 7 & $8.28\pm0.12$ & 6\\

1867: NGC 6715 & $ 4.86\pm0.03 $ & 23 & $4.30\pm0.03$ & 24\\

1868: NGC 6791 & $ 9.78\pm0.09 $ & 9 & $8.77\pm0.09$ & 8\\

1869: NGC 6819 & $ 8.41\pm0.04 $ & 7 & $7.64\pm0.04$ & 7\\

1870: NGC 7078 & $ 2.78\pm0.01 $ & 38 & $2.18\pm0.01$ & 14\\

1871: NGC 7789 & $ 8.47\pm0.02 $ & 20 & $7.61\pm0.02$ & 20\\

1872: \enddata

1873: \end{deluxetable}

1874: 

1875: \end{document}

1876: