Astronomical Science

The Carina Dwarf Spheroidal Galaxy: A Goldmine for Cosmology and Stellar Astrophysics

Peter B. Stetson1 support previous photometric evidence Carina plays a key role among the LG Matteo Monelli2, 3 that both the old (12 Gyr) and the inter- dSph galaxies for many reasons: Michele Fabrizio4 mediate-age (4–6 Gyr) subpopulations i) it is relatively close, and its central Alistair Walker5 show a limited spread in chemical com- ­density is modest; Giuseppe Bono4, 6, 7 position. The use of the Carina galaxy ii) it shows multiple star formation epi­ Roberto Buonanno4, 8 field as a possible standard stellar field sodes, separated in time and in Filippina Caputo6 is outlined. the amount of stellar mass involved Santi Cassisi9 (Monelli et al., 2003); Carlo Corsi6 iii) it hosts a broad variety of variable Massimo Dall’Ora10 Introduction stars ranging from low- and intermedi­ Scilla Degl’Innocenti11, 12 ate-mass hydrogen-burning dwarf Patrick François13 Dwarf galaxies play a fundamental role Cepheids to low- and intermediate- Ivan Ferraro6 in several astrophysical problems. Current mass helium-burning stars (Anomalous Roberto Gilmozzi7 cosmological simulations predict dwarf Cepheids; Dall’Ora et al., 2003); Giacinto Iannicola6 satellite populations significantly larger iv) high resolution spectra are available Thibault Merle14 than the number of dwarfs observed near for fewer than a dozen bright red giants Mario Nonino15 giant spirals like the Milky Way and M31. (RGs). The mean metallicity is [Fe/H] = Adriano Pietrinferni9 This discrepancy is known as the missing –1.69 with a spread of 0.5 (Koch et al., Pier Prada Moroni11, 12 satellite problem and challenges the 2008); Luigi Pulone6 ­current most popular cosmological model: v) calcium triplet measurements based Martino Romaniello7 the Lambda Cold Dark Matter paradigm on medium resolution spectra are also Frederic Thévenin14 (Madau et al., 2008). However, it is not available for a large sample (437) of yet known whether this discrepancy is RG stars (Koch et al., 2006) with a due to limitations in the theoretical mod­ peak at [Fe/H] = –1.72, but ranges from 1 DAO/HIA/NRC, Victoria, Canada elling or to observational bias at the –2.5 to –0.5. Using the same spectra, 2 IAC, Tenerife, Spain faint end of the galaxy luminosity function but different selection criteria, Helmi 3 Univ. La Laguna, Tenerife, Spain (Kravtsov, 2010). et al. (2006) found a metallicity distri­ 4 Univ. Tor Vergata, Roma, Italy bution from about –2.3 to –1.3. 5 NOAO/CTIO, La Serena, Chile The distribution of dwarf galaxies in 6 INAF–OAR, Monte Porzio Catone, Italy multi-dimensional parameter space indi­ However, we are facing a stark discrep­ 7 ESO cates that they follow very tight relations. ancy between the spread in metallicity 8 ASDC. Frascati, Italy In particular, Prada & Burkert (2002) suggested by spectroscopic measure­ 9 INAF–OACTe, Teramo, Italy ­suggested that Local Group (LG) dwarf ments and photometric metallicity indica­ 10 INAF–OAC, Napoli, Italy galaxies show strong correlations (Fun­ tors. Deep and accurate colour–magni­ 11 Univ. Pisa, Pisa, Italy damental Line [FL]) between mass-to- tude diagrams (CMDs) of Carina indicate 12 INFN, Pisa, Italy light (M/L) ratio, surface brightness and that the width in colour of the red giant 13 Obs. de Paris-Meudon, Paris, France metallicity (Z). They explained the corre­­ branch (RGB) is quite limited (Monelli 14 Obs. Côte d’Azur, Nice, France lation between M/L ratio and Z using a et al., 2003; ­Harbeck et al., 2001). This 15 INAF–OAT, Trieste, Italy simple chemical enrichment model where evidence was further supported in a the hot metal-enriched gas, at the end recent inves­tigation by Bono et al. (2010) of the star formation epoch, is lost via employing a large number of multiband We present deep and precise multiband galactic winds (Mac Low & Ferrara, 1999; (U, B, V, I) CCD images from ground- (U, B, V, I) optical data for the Carina Woo et al., 2008). Furthermore, the FL of based telescopes covering the entire dwarf spheroidal galaxy. Data were col- dwarf irregulars (dIs) in a five-dimensional body of the galaxy. lected using three different ground- parameter space (total mass, surface based telescopes (4-metre Blanco brightness, rotation velocity, metallicity CTIO, MPG/ESO 2.2-metre, 1.5-metre and star formation rate) is linear and tight, Photometric datasets and photometric CTIO) and cover the entire body of and extends the scaling relations of calibration the galaxy. We discuss the reliability of giant late-type galaxies to lower mass. the absolute photometric zero-point On the other hand, the FL of dwarf sphe­ The data were collected in various calibrations of images collected with roidals (dSphs) in a four-dimensional observing runs between December 1992 large-format and mosaic CCD cameras. parameter space (total mass, surface and January 2005. They include images The typical precision for the B- , V- brightness, rotation velocity and Z) is also from three telescopes: the Cerro Tololo and I-bands is better than 0.01 mag. linear and very tight, but is not an ex­­tra­ Inter-American Observatory (CTIO) The U-band is a little worse. A prelimi- polation of the scaling relations of giant ­1.5-metre telescope with a single Tektro­ nary comparison with scaled-solar early-type galaxies. nix 2K CCD, the CTIO 4-metre Blanco and α-enhanced evolutionary predic- ­telescope with the Mosaic ii camera, and tions covering a broad range of ages the MPG/ESO 2.2-metre telescope with and chemical compositions seems to the Wide Field Imager (WFI) camera (both

32 The Messenger 144 – June 2011 proprietary and archival data). The data nificant part of his career attempting to repeatability of the observations. For our discussed here represent 4152 individual ensure that measurements of very faint measurements we were able to impose CCD images with essentially complete stars obtained with CCDs on large tele­ the additional condition that each star coverage of the central regions of Carina scopes are on the same photometric shows no evidence of intrinsic variation at (about 40 by 55 arcminutes, lacking only ­system as measurements of very bright a level greater than 0.05 mag root mean those regions obliterated by bright fore­ stars made with photo­multipliers on small square (rms), when the data from all filters ground stars). They were obtained in four telescopes (Stetson, 2000; 2005)1. are considered. photometric bands (B and I with the ­1.5-metre telescope, and UBVI with both The question of whether the photometry The left panels of Figure 1 show the mag­ the 4-metre and 2.2-metre telescopes). discussed here is on the standard system nitude residuals (in the sense published The resulting colour image is shown on breaks down into two parts: minus ours) plotted versus apparent p. 20. i) the extent to which the CCD photome­ ­visual magnitude, while the right panels try, in general, is on the standard sys­ of Figure 1 plot the same residuals The photometric data were reduced tem; and against the B–I colour. The differences using the DAOPHOT/ALLFRAME package ii) the extent to which the photometry for show a systematic departure of a few (Stetson, 1987; 1994). Because of the Carina, in particular, is on the same hundredths of a magnitude between our multiple chips and pointings, any given system as the rest of the photometry. photometric system and the standard star may have up to 17 calibrated meas­ As of now, 2076 datasets have been system for stars brighter than approxi­ urements in U, 156 in B, 207 in V, and homogeneously analysed through our mately V = 9. We infer that the stars with 70 in I. A total of 205 338 individual stars software; among these, 1591 are V ≤ 9 were simply too bright to be meas­ were catalogued and measured. Among ­considered to be of photometric qual­ ured with our equipment, and we were these, 72 595 have photometric measure­ ity, and 485 have been analysed in unduly optimistic about these measure­ ments in all four filters, while 129 230 non-photometric mode (viz. data taken ments. We therefore advise readers to have at least V and either B–V or V–I. The under non-photometric conditions). ignore Stetson’s results for standard stars remainder, either extremely faint (i.e., with V ≤ 9 and rely entirely upon the detectable in only one filter) or located Figure 1 shows the magnitude differ­ ­previously published photometric indices near the periphery, have astrometry and ences between the calibrated standard- of such stars instead. Figure 1 demon­ instrumental magnitudes only. system magnitudes from our own ob­­ strates that our photometry matches the servations of the fundamental photometric definitive published results with a high The fundamental standard system of standards, as compared to the pub- degree of fidelity over a dynamic range of photometric magnitudes was established lished values for the same stars. In order ~600 in flux (9 <~ V <~16) and over the decades ago using photoelectric pho­ to restrict the comparison to stars full available range of stellar temperature. tometers on small telescopes — mostly with the most reliable data, we consider Quantitatively, the weighted mean dif­ of apertures 0.4–0.9 metres (e.g., Land­ only magnitudes based on at least five ferences between the published results olt, 1973; 1988; 1992; Graham, 1982; and independent measurements obtained on and ours are +0.0012 ± 0.0028, +0.0016 references therein to the establishment photometric occasions, and having ± 0.0013, −0.0007 ± 0.0009 and −0.0025 of UBV by Johnson and RI by Cousins). standard errors of the mean not greater ± 0.0013 mag in U, B, V and I, respec­ One of the authors (PBS) has spent a sig­ than 0.02 mag based on the internal tively, based upon 109, 209, 220 and

Figure 1. Left: From top 0.1 0.1 to bottom, difference in U U 0.0 0.0 U-, B-, V- and I-bands δ δ between standard sys­ –0.1 –0.1 tem magnitudes and the current observations 0.1 0.1 of the fundamental ­photometric standards B B 0.0 0.0 δ δ as a function of the vis­ –0.1 –0.1 ual magnitude. Right: Same as the left, but the difference in magnitude 0.1 0.1 is plotted as a function V V δ δ 0.0 0.0 of B–I. –0.1 –0.1

0.1 0.1 I I δ δ 0.0 0.0 –0.1 –0.1

8 10 12 14 16 0.02.0 4.0 6.0 V B–I

The Messenger 144 – June 2011 33 Astronomical Science Stetson P. et al., The Carina Dwarf Spheroidal Galaxy

Figure 2. Left: Same as 0.1 0.1 the left panels of Figure

U 1, but the difference is 0.0 U

δ 0.0 δ between magnitudes –0.1 –0.1 based on datasets from nights when Carina was 0.1 0.1 observed (175 photo­ metric, 88 non-photo­ B 0.0 B δ 0.0 δ metric) as compared –0.1 –0.1 to results from the rest of our nights — when Carina was not 0.1 0.1 observed (1416 clear, V V

δ 0.0 397 cloudy) — for those

δ 0.0 –0.1 –0.1 stars in all our other fields that appear in both datasets. See text 0.1 0.1 for more details. Right: I

I Same as left, but differ­ δ 0.0 δ 0.0 ence in magnitude is –0.1 –0.1 plotted as a function of B–I. 81012 14 16 18 20 22 24 0.02.0 4.0 V B–I

151 stars. Expressed as a standard devi­ our CCD data except in the U filter. In this tude differences are +0.0010, –0.0002, ation, the average standard residual of filter only, 451 stars common to the two –0.0001 mag in B, V and I, respectively, one star is 0.027, 0.017, 0.012, 0.012 and samples — all of which lie in only one based on 8795, 14 945, and 10 329 stars. 0.016 mag in U, B, V, R and I. It does field, Selected Area 98 — show a sys­ The formal statistical confidence intervals not seem that these numbers can be tematic offset of 0.036 mag. on these numbers are all < 1 × 10–4 mag, reduced by obtaining more observations; but these are certainly underestimated; rather, they represent a fundamental The Johnson U bandpass has always we still need to think harder about how to limit to the reliability with which any given been a little problematic. Its classical characterise our actual systematic un­ individual star can be referred to a fun­ implementation provided a bandpass certainties, given that so few independent damental standard system, due to the defined by filter throughput on the long points of comparison are available to us. fact that even though stars may have the wavelength side, but by atmospheric However, the current evidence suggests same broadband colours in one photo­ opacity on the short wavelength side. that systematic differences between our metric system, they can still have underly­ Furthermore, the atmospheric extinction Carina observations and the system of ing spectral energy distributions that in the U-band is very high: ~0.5 mag Landolt and Graham in B, V and I proba­ ­differ in detail, and that will be sampled per airmass on good nights at 2000 bly do not greatly exceed a few times differently by other filter sets. metres on mountaintops like Cerro Tololo 0.001 mag, and calibration issues will not or La Silla. Finally, this bandpass con- be a significant limiting factor for our Figure 2 provides the same sort of com­ tains the Balmer convergence and jump interpretations so long as we use U only parison for all stars in all fields in the in hot stars, and a multitude of strong for differential comparisons within our Carina observations (263 in number: 175 metal lines (including Ca ii H+K) and data. photometric, 88 non-photometric) as bands of molecules like CN in cool stars. compared to results for the same target All of these features make the U magni­ stars appearing in our 1813 datasets tudes sensitive to the exact placement Comparison between theory and obtained on nights when Carina was not and shape of the instrumental bandpass. observations observed (1416 clear, 397 cloudy). Again We will continue to investigate the cause we have restricted the comparison to of this discrepancy, but for the present A detailed comparison between stellar stars with at least five photometric obser­ we expect that a correction of order one- populations in Carina with simple stellar vations, standard errors no larger than tenth to one-half the difference should populations belonging either to Galactic 0.02 mag and no evidence of intrinsic be applied to our U-band results for globular clusters (GCs) or to interme­diate- variation in excess of 0.05 mag rms when Carina to bring them closer to the stand­ age clusters in the Small Magellanic data from all filters are considered. These ard system. Cloud was presented in Bono et al. new comparisons show a few signifi- (2010). The discrimination between field cant photometric outliers with magnitude The problem is that we do not have and Carina stars was performed using differences > 0.10 mag, as expected in many other nights of southern hemi­ the U–V, B–I colour–colour plane. Fig­ any large body of raw data, but the vast sphere observations that included meas­ ure 3 shows three different CMDs of the majority of stars show that the datasets urements in the U-band, to which we candidate Carina stars. Note that after that do include Carina observations are could compare. Quantitatively, apart from the selection there remain some field reasonably well matched to the rest of U, the formal weighted mean magni- stars with colours similar to Carina’s RGB

34 The Messenger 144 – June 2011 (V ≤ 21.5, B–V ~ 0. 5 mag). The evolu­ 16 tionary properties of low- and intermedi­ Carina dSph ate-mass stars provide at least three independent metallicity indicators: the slope and the spread in colour of RGB stars; the spread in colour of red clump 18 (RC) stars; and the spread in magnitude of blue and red horizontal branch (HB) stars. Data plotted in this figure show that both RGB and RC stars cover a very lim­ 20 RC RGB ) HB

ited range in colour in the three adopted ag (m

CMDs. The same observation applies V to the spread in magnitude of blue and red HB stars. 22 The theoretical framework for both scaled-solar and α-enhanced evolu­ IntSGB tionary models is based on isochrones available in the BaSTI database ­(Pietrinferni et al., 2004; 2006). To com­ 24 OLdSGB pare theory and observations we adopted the distance modulus and the 0.0 0.5 1.01.5 012301234 reddening available in the literature B–V (mag) B–I (mag) U–I (mag) (Dall’Ora et al., 2003; Monelli et al., 2003). As a first working hypothesis, we are Figure 3. Left: B–V, V colour–magnitude diagram of stellar tracers (SGB, HB). The red bars plotted at assuming that both the old and the inter­ candidate Carina stars based on optical images the right display the intrinsic photometric error both ­collected with the MOSAIC camera at the 4-metre in magnitude and in colour. Middle: Same as left, mediate-age subpopulations have Blanco telescope, with the WFI at the MPG/ESO but candidate Carina stars are plotted in the B–I, V scaled-solar chemical compositions. ­2.2-metre telescope and with the 1.5-metre CTIO tel­ colour–magnitude diagram. Right: Same as left, Spectroscopic investigations indicate that escope. The blue labels mark intermediate-age but candidate Carina stars are plotted in the U–I, V α-elements in dSph galaxies may be ­stellar tracers (SGB, RC), while the red ones the old colour–magnitude diagram. overabundant when compared with solar composition, but underabundant when isochrones is systematically larger than the old and the intermediate-age predic­ compared with GCs of similar metallicity observed over the magnitude range tions attain colours that are similar to the (Tolstoy et al., 2009; and references of sub-giant branch (SGB) and RGB observed ones. In addition, we repeated therein). stars. The same outcome applies to the the comparison using isochrones assum­ intermediate-age (6 Gyr) subpopulation. ing different values of global metallicity The left panels of Figure 4 show the com­ The current predictions with the most ([M/H], see labelled values). The right parison in the V, B–V CMD, at fixed metal-poor composition attain colours panels of Figure 5 show the comparison chemical composition ([Fe/H] = –1.79, that are similar to the red tail of RC stars. with isochrones of 12 Gyr (top) for the old Y = 0.245), with two old (top; 10, 12 Gyr) However, the observed CMDs show a and of 6 Gyr for the intermediate-age and three intermediate-age (bottom; well defined ­separation between RC and sub-population. 2, 4, 6 Gyr) isochrones. Data plotted in RGB stars. these panels display that theory and The key advantage in the approach we observations, within the errors, agree To further constrain the impact of chemi­ adopted to compare theory and observa­ quite well. The intermediate-age isochro­ cal composition on the evolutionary tions is that we are using theoretical pre­ nes appear slightly bluer than the bulk ­properties of Carina stellar populations, dictions to constrain the relative differ­ of the RGB stars, but the difference is of we assumed that both the old and the ence and the spread in the heavy the order of a few hundredths of a mag­ intermediate-age sub-populations are element abundances of Carina subpopu­ nitude. The right panels of Figure 4 show α-enhanced. The left panels of Figure 5 lations. The above findings indicate that the comparison at fixed age and different show the comparison with two old the possible abundance spreads of both chemical compositions. In particular, (top; 10, 12 Gyr) and three intermediate- the old and the intermediate-age sub­ the top right panel shows the comparison age (bottom; 2, 4, 6 Gyr) isochrones populations appear to be limited and of with three old isochrones of 12 Gyr and ­constructed assuming the same iron the order of 0.2–0.3 dex (Bono et al., iron abundances ranging from –2.3 to content as the scaled-solar isochrones 2010; Fabrizio et al., 2011), thus support­ –1.5 dex. Note that evolutionary models adopted in Figure 4, but an overabun­ ing the spread in iron abundances based were constructed assuming a helium-to- dance of α-elements: [α/Fe] = 0.4. The on high resolution spectra provided by metal enrichment ratio ΔY/ΔZ = 1.4 and a comparison between theory and obser­ Shetrone et al. (2003). ­Theoretical mass-loss rate η = 0.4 (Pietrinferni et al., vation appears slightly better than for and empirical uncertainties do not allow 2004). The colour range covered by the the scaled-solar isochrones, since both us to reach firm conclusions regarding

The Messenger 144 – June 2011 35 Astronomical Science Stetson P. et al., The Carina Dwarf Spheroidal Galaxy

16 and the interplay between Big Bang [Fe/H] = –1.79 dex t = 12 Gyr nucleosynthesis and the creation of Y = 0.245 chemical elements in stars is one of the landmark achievements in astrophy- [Fe/H] = – 2.27 dex sics. The former largely relies on photom­ [Fe/H] = –1.79 dex 18 10 Gyr etry, the latter on spectroscopy. Many 12 Gyr [Fe/H] = –1.49 dex of the current open problems in stellar astrophysics are approached through a comparison between the theoretical and observational Hertzsprung–Russell 20 diagrams. This comparison relies on g) three fundamental pillars: absolute dis­ ma

( tances, photometric calibrations, and V input ­physics adopted in the stellar evolu­ tionary and pulsation models. 22 There is a threefold motivation for the ongoing effort for precise photometric calibrations for stellar populations in nearby dwarf galaxies: μ = 20.11 i) A systematic drift in photometric pre­ 24 E(B–V) = 0.03 cision when moving from bright to faint RG stars might produce a spread in 16 colour, and in turn, mimic a spread in metal content. Moreover, these sys­ t = 6 Gyr tems are characterised by very low stel­ lar densities, thus requiring that pho­ 2 Gyr tometry be collected with large format 18 4 Gyr and mosaic CCD cameras. Intense 6 Gyr effort to achieve consistent zero-points across the field is mandatory to avoid spurious broadening of the key evolu­ tionary sequences. ii) Dwarf galaxies in the Local Group and 20 g) in the local volume are often con­

ma taminated by foreground field stars. (

V Their colours, unfortunately, are similar to the colours of Galactic RGs and main sequence (MS) stars, once again 22 mimicking a broadening in colour due to a spread in age and/or in metal­ licity. The separation of field and Galactic stars can be performed using radial velocity and/or proper motions, 24 but at present this approach is lim- ited to ­relatively bright RGs. A robust cleaning of MS and SG sequences can 0.0 0.51.0 1.5 0.0 0.51.0 1.5 be ­performed using colour–colour B–V (mag) B–V (mag) planes (Bono et al., 2010) and/or a sta­ tistical subtraction of a control field Figure 4. Left: B–V, V colour–magnitude diagram of the possible occurrence of an α-element (Balbinot et al., 2010). These ap­­ Carina based on optical images. The solid coloured enhancement. proaches require, once again, very lines display two old (top) and three intermediate-age (bottom) scaled-solar isochrones at fixed iron abun­ precise calibration in different bands dance and different ages. The adopted true distance across multiple telescope pointings. modulus and the reddening are also indicated. Right: Future prospects iii) The use of variable stars as distance Same as left panels, but comparison assumed an indicators, stellar tracers and phys- age of 12 Gyr for the old (top) and of 6 Gyr for the intermediate-age (bottom) subpopulation. The identification of the physical mecha­ ics laboratories requires precise nisms that drive the formation and time series data and multiband cali­ the evolution of stars and stellar systems, brations.

36 The Messenger 144 – June 2011 16 and in near-infrared (NIR) bands. All three [Fe/H] = –1.84 dex t = 12 Gyr Carina fields have 659 U-band, 5711 B- [α/Fe] = +0.4 dex and V-band and 1467 I-band standard [M/H] = +0.4 dex stars spanning a region 123 by 149 arc­ [Fe/H] = – 2.14 dex minutes. The brightest standard has Y = 0.246 [Fe/H] = –1.84 dex 18 V = 13.95; the median is V = 19.27; and [Fe/H] = –1.62 dex the faintest V = 22.42 mag. The Carina region together with other regions1 is thus 10 Gyr a potential standard stellar field to cali­ 12 Gyr brate the next generation of large CCD 20 cameras. g) ma

( The use of accurate and deep NIR data, V together with the present dataset, can play a crucial role in the identification of field and Carina candidate stars using 22 optical–NIR colour–colour planes. The stronger sensitivity to effective tempera­ ture of the optical–NIR colours can also provide a robust separation between μ = 20.11 old and intermediate-age stars. At the 24 E(B–V) = 0.03 same time, a high-resolution, high signal- to-noise ratio spectroscopic study of 16 a sizable sample of RG stars and He- burning (RC, HB) stars is mandatory to t = 6 Gyr measure the abundance spread, if any, in old and intermediate-age stellar popu­ 2 Gyr (M/H) = 1.79 dex lations. 18 4 Gyr (M/H) = 1.49 dex 6 Gyr (M/H) = 1. 29 dex References

Balbinot, E. et al. 2010, MNRAS, 404, 1625 Bono, G. et al. 2010, PASP, 122, 651 Dall’Ora, M. et al. 2003, AJ, 126, 197 20

g) Fabrizio, M. et al. 2011, PASP accepted Graham, J. A. 1982, PASP, 94, 244 ma

( Helmi, A. et al. 2006, ApJL, 651, L121 V Koch, A. et al. 2006, AJ, 131, 895 Koch, A. et al. 2008, AJ, 135, 1580 Kravtsov, A. 2010, Advances in Astronomy, 2010, 22 Landolt, A. U. 1973, AJ, 78, 959 Landolt, A. U. 1983, AJ, 88, 439 Landolt, A. U. 1992, AJ, 104, 340 Mac Low, M. & Ferrara, A. 1999, ApJ, 513, 142 Madau, P. et al. 2008, ApJ, 679, 1260 Monelli, M. et al. 2003, AJ, 126, 218 24 Pietrinferni, A. et al. 2004, ApJ, 612, 168 Pietrinferni, A. et al. 2006, ApJ, 642, 797 Prada, F. & Burkert, A. 2002, ApJ, 564L, 73 Shetrone M. D., et al. 2003, AJ, 125, 684 Stetson, P. B. 1987, PASP, 99, 191 0.0 0.51.0 1.5 0.0 0.51.0 1.5 Stetson, P. B. 1994, PASP, 106, 250 B–V (mag) B–V (mag) Stetson, P. B. 2000, PASP, 112, 925 Stetson, P. B. 2005, PASP, 117, 563 Figure 5. Left: Same as left panels of Figure 4, but The way to easily overcome the above Tolstoy, E., Hill, V. & Tosi, M. 2009, ARA&A, 47, 371 comparison performed using isochrones con­ problems is to have a sizable sample Woo, J., Courteau, S. & Dekel, A. 2008, MNRAS, structed by assuming the same iron content 390, 1453 ([Fe/H] = –1.84 dex) of the scaled-solar models, but of local standard stars covering a broad with an α-enhanced chemical composition (see range of magnitudes and colours. labelled values). Right: Same as left panels, but Local standards are now available in a Links comparison performed using isochrones with differ­ significant fraction of globular clusters. ent global metallicities ([M/H], see labelled values). 1 List of photometric standard fields: http://www3. So far, only a few dwarf galaxies have cadc-ccda.hia-iha.nrc-cnrc.gc.ca/community/ precise local standard stars in the optical STETSON/standards/

The Messenger 144 – June 2011 37