A Keck/DEIMOS spectroscopic survey of the faint M31 satellites And XV and And XVI
B. Letarte, S. C. Chapman, M. Collins, R. A. Ibata, M. J. Irwin, A. M. N. Ferguson, G. F. Lewis, N. Martin, A. McConnachie, N. Tanvir
aa r X i v : . [ a s t r o - ph . S R ] A ug Mon. Not. R. Astron. Soc. , 000–000 (0000) Printed 19 October 2018 (MN L A TEX style file v2.2)
A Keck/DEIMOS spectroscopic survey of the faint M 31satellites And XV and And XVI
B. Letarte , S. C. Chapman , M. Collins , R. A. Ibata , M. J. Irwin ,A. M. N. Ferguson , G. F. Lewis , N. Martin , A. McConnachie , N. Tanvir California Institute of Technology, 1200 E. California Blvd, MC 105-24, Pasadena, CA 91125 USA Institute of Astronomy, Madingley Road, Cambridge, CB3 0HA, U.K. Observatoire de Strasbourg, 11, rue de l’Université, F-67000, Strasbourg, France Institute for Astronomy, University of Edinburgh, Royal Observatory, Blackford Hil l, Edinburgh, UK EH9 3HJ Institute of Astronomy, School of Physics, A29, University of Sydney, NSW 2006, Australia Max-Planck-Institut fur Astronomie, Konigstuhl 17, D-69117 Heidelberg, Germany Department of Physics and Astronomy, University of Victoria, Victoria, B.C., V8P 1A1, Canada Department of Physics & Astronomy, University of Leicester, Leicester, LE17RH, UK
19 October 2018
ABSTRACT
We present the results of a spectroscopic survey of the recently discovered M31 satel-lites And XV and And XVI, lying at projected distances from the centre of M31 of 93and 130 kpc respectively. These satellites lie to the South of M31, in regions of thestellar halo which wide field imaging has revealed as relative voids (compared to the ∼ degree-scale coherent stream-like structures). Using the DEep Imaging Multi-ObjectSpectrograph mounted on the Keck II telescope, we have defined probable membersof these satellites, for which we derive radial velocities as precise as ∼ − down to i ∼ ±
50 kpc), we have demonstrated that the brightest three stars pre-viously used to define the tip of the red giant branch (TRGB) in And XV are infact Galactic, and And XV is actually likely to be much more distant at 770 ± V ≈ − . to M V ≈ − . . The And XV velocity dispersion is resolved with v r =-339 +7 − km s − and σ v = 11 +7 − km s − . The And XVI dispersion is not quite resolved at σ withv r =-385 +5 − km s − and σ = 0 +10 − indef km s − . Using the photometry of the confirmedmember stars, we find metallicities of And XV (median [Fe/H] = − . , interquar-tile range ± = − . , interquartile range ± = − . (-2.1)for And XV (And XVI), with a uncertainty of ∼ . dex in both cases. Our measure-ments of And XV reasonably resolve its mass ( ∼ M ⊙ ) and suggest a polar orbit,while the velocity of And XVI suggests it is approaching the M31 escape velocity givenits large M31-centric distance. Key words:
M 31 – Dwarf Galaxies – DEIMOS
In recent years, systematic searches have been performedin earnest within the Local Group, both photometricallyand spectroscopically, in order to fully catalogue its pop-ulation of satellite galaxies. There are several motivationsbehind these endeavours, with one of the more hotly dis-cussed being that of the so-called “missing satellites prob-lem” (Moore et al. 1999; Klypin et al. 1999), which is anobserved discrepancy of 1-2 orders of magnitude between the number of satellite galaxies produced within cosmo-logical simulations and the number we see orbiting theGalaxy and M31 today. And whilst these surveys haveapproximately doubled the known satellite populationsof both galaxies within recent years (Zucker et al. 2004;Willman et al. 2005; Chapman et al. 2005; Belokurov et al.2006a; Martin 2006; Zucker et al. 2006a,b; Belokurov et al.2007a; Majewski 2007; Belokurov et al. 2007a; Irwin et al.2008) we are still far from reconciling the observations with c (cid:13) Bruno Letarte et al. simulations. Several theories have been put forward in or-der to address this issue. Survey completeness is a widelydebated topic (see e.g. Tollerud et al. 2008; Simon & Geha2007), particularly with respect to the MW, where observa-tional efforts are hampered by obscuration from the diskand the bulge. Future all-sky surveys may find a wealthof ultra-faint satellites that would allieviate this gap be-tween observation and theory. Another proposed solutionis that the satellite galaxies within the local group inhabitincreasingly dark matter dominated halos as they becomefainter (Bullock, Kravstov & Weinberg 2000; Stoehr et al.2002), so that even the faintest of dwarf galaxies would re-side within massive dark matter halos.Penarrubia et al. (2006, 2008a, 2008b) demonstrate thatwithin the LCDM working frame, dwarf spheroidal galaxies(dSphs) are well embedded within the dark matter halosthat surround them. For the brightest dwarf galaxies in theMW they find that . < R c /r s < . , where R c and r s are the stellar core radius and the NFW scale radius, respec-tively. That implies that these systems are fairly resilient totidal interactions, so that they can lose a large fraction oftheir initial dark matter mass before the stellar componentstarts being disrupted.An observational test of these suggestions is then tomeasure the stellar light and dynamical mass of faint dwarfgalaxies in the Local Group. With spectroscopic informa-tion, one can measure the central velocity dispersion ofthe satellite, which has been shown to be a good indica-tor of the instantaneous mass of a dwarf galaxy withinthe radii of the luminous tracers (e.g. Oh, Lin & Aarseth1995; Piatek & Pryor 1995), even if the dwarf is not invirial equilibrium or perfectly spherical. Several approacheshave been taken to derive the total mass of the systemfrom the central velocity dispersion (e.g. Illingworth 1976;Richstone & Tremaine 1986; Gilmore et al. 2007). Recentdetailed modelling of dwarf galaxy velocity dispersions sug-gests a lower dynamical mass limit of ∼ M ⊙ (Strigari etal. 2008).Spectroscopic data on the dSphs of the local group isalso desirable in order to give us a better understanding ofthe nature of these tiny galaxies. The discovery and studyof the faintest members of this population has shown thatat the low luminosity end of the spectrum, these satellitesdo not behave as expected from the study of their brightercounterparts. In particular, they begin to deviate signifi-cantly from established trends between mass and metallicity,and M/L vs light. In the brighter MW dwarfs, the lack ofvery metal poor stars (Helmi et al. 2006), the stellar popu-lations (Unavane et al. 1996) and differing abundance pat-terns from the MW halo (Shetrone et al. 2001) reinforce thepoint that the surviving, isolated dwarfs have different envi-ronmental conditions than those which phase mixed into theMW stellar halo. However, the ultra-faint MW satellites doshow evidence of very metal poor stars (Kirby et al. 2008),possibly suggesting these dwarfs are remnants of the halobuilding blocks.By aquiring kinematic data for the Local Group satel-lites, we can isolate the likely dwarf member stars from theColour Magnitude Diagrams (CMD), perform independentchecks on the metallicities derived photometrically, and alsoestimate their total masses so we can place them more fullyin the context of the dSph population, and determine if there truly is a breakdown in these relations at low luminosities,and if so, at what point this divergence begins to take effect.Another motivation behind obtaining kinematic datafor the satellite populations of the MW and M31 is toachieve a greater understanding of the dynamics of the Lo-cal Group as a whole. The mass contained within the in-ner few 10’s kpc of both is relatively well constrained fromoptical and HI data, as well as globular cluster and plan-etary nebulae tracer populations (see e.g. Kochanek 1996;Carignan et al. 2006), but probing the outer halo at largeradii proves to be more of a challenge. An ideal method formeasuring the total masses of the MW and M31 is to usetheir satellite galaxies as direct tracers of their potentials. Asthey are located at large distances from their hosts ( ∼ few100’s kpc) they probe the full extent of their mass distri-butions. Such a method has been used previously in boththe MW and M31 (Evans et al. 2000; Wilkinson & Evans1999; Gottesman, Hunter & Boonyasait 2002), however theassociated errors on their measurements are huge (of order ∼ twice the measurements themselves) due to the low num-ber of satellite galaxies with reliable kinematic and distancedata available to them. In the years since these results werepublished, more than 20 satellite companions of the MWand M31 have been discovered, many of which have alsobeen studied spectroscopically. If kinematic data on all thesatellites within the Local Group can be obtained, we willbe able to better constrain the masses of these two “sister”galaxies.With these goals in mind, it remains crucial to obtainradial velocities for newly discovered dSphs in order to un-derstand their mass distribution and orbital properties. Asa step towards this end, we have used the Keck II DEepImaging Multi-Object Spectrograph to derive radial veloci-ties and metallicities of stars within two new satellites dis-covered in Ibata et al. (2007), And XV and And XVI. Multi-object Keck observations with DEIMOS (Faber et al.2003) for And XV and And XVI were made on 2007 Oct 8–9, in variable conditions (with 0.6–1.0 ′′ seeing and patchycirrus). To obtain improved S/N in a reasonable exposuretime for the faint ( i = 20 . − . ) RGB stars targeted,we employed the lower resolution 600 line/mm grating. Forthe Ca II triplet (CaT) lines, which are reasonably resolvedat this lower (R ∼ > ◦ A over the CaTregions for faint ( i ∼ ) stars, the velocity errors can sufferlarger systematic errors when they lie close to OH lines, andthe velocity accuracy can vary considerably with continuumS/N. Our chosen instrumental setting covered the observedwavelength range from . − . µ m. Exposure time was60 min on each dSph, split into 20-min integrations. Datareduction followed standard techniques using the DEIMOS-DEEP2 pipeline (Faber et al. 2003), debiassing, flat-fielding,extracting, wavelength-calibrating and sky-subtracting thespectra.The radial velocities of the stars were then measured by c (cid:13) , 000–000 fitting the peak of the cross-correlation function derived us-ing a template spectrum consisting of delta functions to theinstrumental resolution (3 ◦ A ) at the wavelengths of the CaTabsorption lines. This procedure also provides an estimateof the radial velocity accuracy obtained for each measure-ment. The velocity uncertainties typically range from to
25 km s − . Target stars were assigned from a broad (1 mag)box around the general outline of the RGB of the dwarf.A narrower box of 0.3 mag around the RGB was then con-structed with higher priorities. In both of these selectionboxes, the priority was scaled as a function of i -mag so thatbrighter stars had more chances of being observed. The re-mainder of the mask space was filled with target stars froma broad region encompassing RGB stars in M31 over allpossible metallicities and distances. The And XV, And XVImasks had 143, 131 target stars respectively. The small num-ber of candidate members compared to the total amount ofstars observed is simply due to the fact that the footprint ofDEIMOS is much wider than the small angular size of thedwarf galaxies, therefore, there are many more stars that arenot part of the main dSph targets. The spatial distributions of the stars in the regions ofAnd XV, And XVI are shown in Figure 1. Candidate mem-ber stars from the DEIMOS spectroscopy are highlighted(listed in Tables 1 & 2). Member stars are selected by hav-ing a velocity which lies within σ of the median of theobvious kinematic clumps of stars in each dwarf field. Notethat for And XV, the member stars are highlighted differ-ently for robust and tentative members, defined below interms of the strength of the cross-correlation peak in thevelocity determination. CFHT-MegaCam colour-magnitudediagrams (CMDs) of stars within a three arcminute radiusof And XV and And XVI are shown in Figure 2, again withlikely member stars highlighted. The red giant branches ofboth dwarfs are clearly visible, while the horizontal branchesare reasonably detected. In the case of And XV, three starspreviously assumed to lie near the TRGB have been shownby their velocities and NaI doublet equivalent width (EW)to be foreground Galactic dwarf stars. We then revise thedistance estimate of And XV from the 630 ±
60 kpc previ-ously reported (Ibata et al. 2007), estimating the true tip ofthe RGB is ∼ i = 20 . previouslyassumed. Adopting MTRGB = -4.04 ± I -band magnitudeof the RGB tip, and convert into the Landolt system usingthe colour equations in Ibata et al. (2007) and those given byMcConnachie et al. (2004); this yields a distance modulusof m − M = 24 . ± . or a distance of 770 ±
70 kpc (ver-sus the previous 630 ±
60 kpc). This revised distance agreeswith that obtained by considering the magnitude, g ∼ . ,of the Horizontal Branch (Fig. 2). The change in distanceincreases the luminosity from M V ≈ − . to M V ≈ − . . InAnd XVI, as all stars near the TRGB were confirmed spec-troscopically as members of the dwarf spheroidal galaxy, theIbata et al. (2007) distance estimate of 525 ±
50 kpc remainsunchanged. The horizontal branch for And XVI (Fig. 2) at g ∼ . (0.8 mag difference from And XVI) agrees well withour proposed 0.5 mag difference in the TRGB between the two dwarf spheroidals. In Fig. 2, we also overlay 13 Gyr oldPadova isochrones with solar-scaled chemical compositions(Girardi et al. 2004), at the TRGB distance and medianmetallicity obtained from member stars (see below), provid-ing a reasonable fit in both cases. The use of an old (13 Gyr)isochrone can be justified by the predominantly old natureof the majority of the recently detected M31 satellites.The distribution of stars in And XV and And XVI isshown in Fig. 3 as a function of their radial velocity. Thestars are then shown as a function of radius from the cen-ters of the dwarfs with half-light radii indicated. Photomet-ric [Fe/H] is estimated by comparison to Padova isochrones(Girardi et al. 2004) corrected for extinction (Schlegel et al.1998) and shifted to the revised distances of the dSphs.These [Fe/H] values are shown in Fig. 3 as a function oftheir radial velocity, revealing the tight range in metallicitiesof both And XV (median [Fe/H]=-1.58, interquartile range ± ± > . , and more tentative member starswith cross-correlation peak < . , but still lying on thewell defined RGB of And XV. The inverse variance weighted,summed spectrum is shown in the bottom offset panel foreach dSph. All candidate member spectra for And XVI areconsidered to be robust by the above criteria. While notshown in the spectra, the Na I doublet is undetected sig-nificantly in the individual stars, and also in the stackedspectra. The Na I equivalent width is sensitive to the surfacegravity, and thus is a good discriminant of Galactic fore-ground dwarf stars (Schiavon et al. 1997), although at theselarge negative velocities, the probability is very low that anyidentified star in our CMD selection box would be Galactic.Stars with velocities <
150 km s − , consistent with Galacticstars typically have well detected Na I doublet lines in ourDEIMOS spectra. The spectroscopic [Fe/H] has very largeerrors in most of the individual spectra, however a reason-able comparison can be made between the [Fe/H] derivedfrom the stacked spectrum and the median photometric[Fe/H]. We find [Fe/H] = − . (-2.1) for And XV (And XVI)by measuring the EW of the Ca II triplet lines as in Chap-man et al. (2005), on the Carretta & Gratton (1997) scale.A large uncertainty, measured from the summed sky spec-trum, of ∼ . dex is due primarily to sky-subtraction resid-uals making it difficult to define the continuum level of thespectrum. Koch et al. (2008) have also demonstrated thathigh quality Keck/DEIMOS spectra of stars in the M31 haloare amenable to further chemical analysis, showing a rangeof species (mostly Fe I and Ti I lines) which become weakerfor the more metal-poor stars. For And XV and And XVI,even the stacked spectra do not detect significant absorp-tion at the Ti I lines (8378, 8426, and 8435 ◦ A ). Our stackedspectra do not have the requisite S/N to detect the antici-pated EW based on stars of similar metallicity from Kochet al. (2008), although roughly tripling the S/N of the spec-tra would be more than sufficient ( ∼ × integration time c (cid:13) , 000–000 Bruno Letarte et al. in typical Mauna Kea weather conditions, compared to therather poor conditions under which the present data weretaken).For our samples of member stars with large and vari-able velocity errors (typically 6–25 km s − ), the MaximumLikelihood approach (e.g., Martin et al. 2007) provides amethod to assess the true underlying velocity distributionof the dSphs. Figure 6 shows the results of our analysis ofAnd XV and And XVI plotting the one dimensional distri-butions 1, 2, 3 × σ values. The And XV dispersion is resolvedwith v r =-339 +7 − km s − and σ v = 11 +7 − km s − , where littledifference is found from using only the best seven memberstars (from Fig. 4) versus using all 13 candidate members(including the six more tentative velocities measurements).The And XVI dispersion is not quite resolved at σ with v r =-385 +5 − km s − and σ = 0 +10 − indef km s − . We have been able to further constrain the properties of thedSphs And XV, And XVI by measuring velocities and metal-licities and assigning probable member stars to each galaxy.While this represents a significant addition to our knowledgeabout these dSphs compared to the photometric discoveryand characterization (Ibata et al. 2007), our spectroscopicmeasurements are generally not precise enough to providerobust measures of the velocity dispersions. And XV has amost likely dispersion ∼
10 km s − (while And XVI is morepoorly constrained but has a σ upper limit of
10 km s − )which would translate to a ∼ M ⊙ halo mass using themethod of Richstone & Tremaine (1986) as has been donefor other M31 dSphs with better constrained velocity dis-persions (e.g., Chapman et al. 2005, Majewski et al. 2007).This is similar to that proposed by Gilmore et al. (2007)for a limiting dark matter halo mass in the smallest galax-ies, however we cannot confidently rule out that the disper-sions are smaller than our measurements suggest. LongerKeck/DEIMOS exposures under good conditions could sig-nificantly improve these results. The relatively well popu-lated RGBs of both dSphs suggests that enough memberstars could be obtained to begin to measure the velocitydispersion profile, a much more reliable constraint on thedark matter halo mass (Gilmore et al. 2007).As mentioned, both dSphs lie in relative voids (free ofdegree-scale substructure) in the M31 halo maps of Ibataet al. (2007). In our spectroscopic samples there are rela-tively few remaining M31 halo stars once the likely And XV,And XVI member stars and Galactic foreground are removed(3 and 6 candidate halo stars respectively, and some ofthese are likely to be Galactic by their proximity to thelow-velocity regime of Milky Way stars). This is consistentwith the extrapolated M31 halo profle (Ibata et al. 2007)out to 93 kpc and 130 kpc respectively. However, this sug-gests that significant spectroscopic efforts would be requiredto properly characterize the (substructure-free) halo in the100-150 kpc regime. Nonetheless, the average spectroscopic[Fe/H]=-1.5 of these 9 confirmed M31 halo stars is consistentwith a lack of metallicity grandient in the underlying metalpoor ([Fe/H]=-1.4) halo found within the inner 70 kpc ofM31 (Chapman et al. 2006). Spectroscopic studies of RGBstars along the minor axis of M31 have been prone to inad- vertently sampling stream-like substructure (Gilbert et al.2006; Koch et al. 2008; Chapman et al. 2008), but they alsodo find a range of metallicities in the 100-150 kpc regimeconsistent with the [Fe/H] in these dSph fields.We have described before in Chapman et al. (2008) theexpected probabilities of halo star contamination to fieldsat these distances. Here we run a simple Monte Carlo simu-lation of the expected effect of the contaminant on the ob-served σ v . We take two distributions, for the halo (120km/s σ v at 100kpc projected) and for each dSph as measuredhere. We then draw randomly from these distributions, ask-ing how often halo stars land in some window indistin-guishable from the dSph. We then remeasure the dwarf σ v whenever star(s) from the halo are present. We find a halostar lands in the And XV window 25% of the time, and inAnd XVI 9% of the time. In 100 samples with a contami-nating halo star(s) lying in the dSph, the velocity disper-sion was measured to range 11-12 km/s (And XV, assumed σ v =11km/s) and 8-9 km/s (And XVI, assumed σ v =8 km/s).Thus the affect of halo stars lying in the velocity window issmall relative to the errors in estimating σ v .We then turn to the orbital properties of these dSphs.And XV lies near the minor axis of M31, at about the samedistance and heliocentric velocity as M31 (785 kpc and v r ∼ -300 km s − ). As such, its orbit must be close to polar witha large implied tangential velocity component. By contrast,at the large M31-centric distance of And XVI ( ∼
270 kpc),its velocity of ∼ −
400 km s − pushes it towards the escapevelocity of M31 (assuming the mass modeling of Geehan etal. 2006; see Chapman et al. 2007 Fig. 4 for a depiction ofthe M31 dwarf galaxies in this context). This is compara-ble to the orbital properties other recently discovered dSphs(And XII: Chapman et al. 2007; And XIV: Majewski et al.2007). In particular, And XII is traveling near the escape ve-locity of the entire Local Group. Finding so many satellitesnear the escape velocity of M31 may suggest that the totalmass of M31 has been somewhat underestimated to date.And XVI appears to sit spatially within a preferred dis-tribution of satellites of M31 (Koch et al. 2006; McConnachie& Irwin 2006), falling towards us at ∼
100 km s − (withan unknown tangential component). However, And XV sitssomewhat outside of this distribution (in both position andlikely orbit) and may represent one of the growing number ofdiscovered M31 satellites which hint at a more complicatedenvironment than was apparent with the first generation ofdiscovered satellites. The data presented herein were obtained at the W.M. KeckObservatory, which is operated as a scientific partnershipamong the California Institute of Technology, the Univer-sity of California and the National Aeronautics and SpaceAdministration. The Observatory was made possible by thegenerous financial support of the W.M. Keck Foundation.
REFERENCES
Bellazzini, M.; Ferraro, F. R.; Pancino, E., 2001, MNRAS,327L, 15 c (cid:13) , 000–000 Table 1.
Properties of candidate member stars in And XV centered at α = 01h 14m 18.7s, δ = 38 ◦ ′ ” . α (J2000) δ (J2000) v r ( km s − ) [F e/H] spec S/N ctm b [F e/H] phot g -mag i -mag01:14:16.28 38:05:49.6 -360.0 ± ± ± ± ± ± ± ± ± ± ± ± ± a Candidate member stars in And XV are divided into a robust group (top 7) and a more tentative group (bottom 6). b Signal to noise in the continuum, estimated over the 250 ◦ A region surrounding the CaT. Table 2.
Properties of candidate member stars in And XVI centered at α = 0h 59m 30.0s, δ = 32 ◦ ′ ” . α (J2000) δ (J2000) v r ( km s − ) [F e/H] spec S/N ctm [F e/H] phot g -mag i -mag00:59:29.54 32:21:59.9 -373.5 ± ± ± ± ± ± ± ± Bellazzini, M.; Ibata, R. A.; Chapman, S. C.; Mackey, A.D.; Monaco, L.; Irwin, M. J.; Martin, N. F.; Lewis, G. F.;Dalessandro, E. 2008, AJ, 136, 1147Belokurov V. et al., 2006, ApJ 647, L111Belokurov V. et al., 2007a, ApJ 658, 337Belokurov V. et al., 2007b, ApJ„ 654, 897Belokurov V. et al., 2008, ApJ, 686L, 83Bullock J. S., Kravstov A. V. & Weinberg D. H. 2000, ApJ539, 517Carignan C., Chemin L., Huchtmeier W. K., Lockman F.J. 2006, ApJ, 641L, 109Carretta, E., Gratton, R. 1997, A&AS 121, 95Chapman S., Ibata R., Ferguson A. M. N., Lewis G., IrwinM. & Tanvir N. 2005, ApJ 632, L87Chapman S., Ibata R., Lewis G., Ferguson A. M. N., IrwinM., McConnachie A., Tanvir N. 2006, ApJ 653, 255Chapman S., Penarrubia J., Ibata R., McConnachie A.,Martin N., Irwin M., et al., 2007, ApJ 662, 79L Mc-Connachie, A. W., PeŰarrubia, J., Navarro, J. F., 2007,Chapman, S. C.; Ibata, R.; Irwin, M.; Koch, A.; Letarte,B., et al., 2008, MNRAS, 390, 1437Evans, N. .; Wilkinson, M. I., Guhathakurta, P, Grebel, E.K., Vogt, S., 2000, ApJ, 540L 9Faber, S. M., et al., 2003, SPIE, 4841, 1657Fellhauer M., et al., 2007, MNRAS, 375, 1171Geehan J., Fardal M., Babul A., Guhathakurta P. 2006, MNRAS, 366, 996Girardi, L., Grebel, E., Odenkirchen, M., Chiosi, C. 2004,A&A, 422, 205Gilbert K., et al. 2006 ApJ, 652, 1188Gilmore G.., et al. 2007, NuPhS, 173, 15Gottesman S. T., Hunter J. H., Boonyasait V. 2002, MN-RAS, 337, 34Helmi, A., et al., 2006, ApJ, 651, 121LHuxor A. et al. 2005, MNRAS, 360, 1007Huxor A. et al. 2008, MNRAS, 385, 1989Ibata R., Chapman S., Ferguson A., Lewis G., Irwin M.,McConnachie A., Tanvir N., 2004, ApJIbata R., Chapman S., Ferguson A., Lewis G., Irwin M.,McConnachie A., Tanvir N., 2005, ApJ 634, 287Ibata R., Chapman S., Irwin M., Lewis G. & Martin N.,2006, MNRAS 373, 70LIbata, R.; Martin, N. F.; Irwin, M.; Chapman, S.; Ferguson,A. M. N.; Lewis, G. F.; McConnachie, A. W.2007, ApJ671, 1591Illingworth G., 1976, ApJ 204, 73Irwin, M. J.; Ferguson, A. M. N.; Huxor, A. P.; Tanvir, N.R.; Ibata, R. A.; Lewis, G. F.Kirby, E. N.; Simon, J. D.; Geha, M.; Guhathakurta, P.;Frebel, A. 2008, ApJ, 685L, 43Klypin A., Kravstov A. V., Velenzuela O. & Prada F. 1999,ApJ 522, 82 c (cid:13) , 000–000 Bruno Letarte et al.
Figure 1.
Spatial distribution of stars in the regions of And XV(top) and And XVI bottom Candidate member stars identifiedspectroscopically are highlighted as robust (filled squares) likely(filled triangles), and in the case of And XV three Galactic fore-ground stars which were previously identified as the tip of theRGB (pentagons). Circular half-light radii are overlaid.
Koch A., Grebel E. K., Wyse F. G. W., Kleyna J. T.,Wilkinson M. I., Harbeck D. R., Gilmore G. F. & EvansN. W., 2006, AJ 131, 895Koch, A.; Rich, R. M.; Reitzel, D. B.; Martin, N. F.; Ibata,R. A.; Chapman, S. C. et al.2008 ApJ, 689, 958Kochanek, C. S., 1996, ApJ, 457, 228Lewis, G. F.; Ibata, R. A.; Chapman, S. C.; McConnachie,A.; Irwin, M. J.; Tolstoy, E.; Tanvir, N. R. 2007, MNRAS, 375, 1364Majewski S., et al., 2007, ApJ, 670, 9LMartin N. F., Ibata R. A., Irwin M. J., Chapman S.C., Lewis G. F., Ferguson A. M. N., Tanvir N. & Mc-Connachie A. W., 2006, MNRAS371, 1983Martin N. F., Ibata R. A., Chapman S. C., Irwin M. J.,Lewis G. F., 2007, MNRAS380, 281McConnachie, A., Irwin, M., Ferguson, A., Ibata, R., Lewis,G., Tanvir, N., 2004, MNRAS 350, 243McConnachie, A., Irwin, M., Ferguson, A., Ibata, R., Lewis,G. Tanvir, N. 2005, MNRAS 356, 979McConnachie, A. W., Peñarrubia, J., Navarro, J. F., 2007a,MNRAS, 380, L75McConnachie, A. W., Arimoto, N., Irwin, M., 2007, MN-RAS, 379, 379McConnachie, A. W.; Huxor, A.; Martin, N. F.; Irwin, M.J.; Chapman, S. C., et al. 2008 ApJ, 688, 1009Moore B., Ghigna S., Governato F., Lake G., Quinn T.,Stadel J. & Tozzi P. 1999, ApJ 524, L19Oh K. S., Lin D. N. C. & Aarseth S. J. 1995, ApJ 442, 142Peñarrubia J., McConnachie A., Babul A., 2006, ApJ, 650,33Peñarrubia J., McConnachie A., Navarro J., 2008a, ApJ,672, 904Peñarrubia J., Navarro J., McConnachie A., 2008b, ApJ,673, 226Piatek S.& Pryor C. 1995, AJ 109, 1071Richstone D. & Tremaine S., 1986, AJ 92, 72Shetrone M., Cote P., Sargent W., 2001, ApJ, 548, 592Schiavon R. P., Barbuy B., Rossi S. C. F. & Milone A.,1997, ApJ 479, 902Schlegel, D., Finkbeiner, D., Davis, M. 1998, ApJ 500, 525Simon J., Geha M., ApJ, in press (2007arXiv0706.0516)Stoehr F., White S. D. M., Tormen G. & Springel V. 2002,MNRAS 335, L84Strigari L., Bullock J., Kaplinghat M., Simon J., Geha M.,Willman B., Walker M. G. 2008 Nature, .454, 1096Tollerud E. J., Bullock J. S., Strigari L. E., Willman B.2008, ApJ 688Tolstoy E. et al. 2004, ApJ 617, L119Unavane M., Wyse R., Gilmore G., 1996, MNRAS, 278, 727Wilkinson M. I., Evans, N. W., 1999, MNRAS, 310, 645Willman B. et al. 2005, AJ 129, 2692Zucker et al. 2006a, ApJ 643, L103Willman B. et al. 2006, AJ submitted, astro-ph/0603486Zucker et al. 2006a, ApJ 643, L103Zucker et al. 2006c, ApJ astro-ph/0606633Zucker et al. 2007, ApJ 612, L121 c (cid:13) , 000–000 Figure 2.
CFHT-MegaCam colour-magnitude diagrams of starswithin a three arcminute radius of And XV (top) and And XVI(bottom). The MegaCam mags are calibrated on SDSS-like sys-tem as AB mags. The red giant branches of both dSphs are clearlyvisible, while the horizontal branch is reasonably detected in bothcases. Candidate member stars from the DEIMOS spectroscopyare highlighted (filled squares – robust members, filled triangles– tentative members). In the case of And XV, three stars previ-ously assumed to lie near the TRGB have been shown by theirvelocities and Na I doublet EWs to be Galactic foreground (openpentagons). The revised TRGB and horizontal branch distanceto And XV is 770 kpc as described in the text. 13 Gyr old Padovaisochrones are overlaid, at the TRGB distance and median metal-licity obtained from member stars, providing a reasonable fit inboth cases. Figure 3.
Distribution of stars in And XV (top), And XVI (bot-tom) as a function of their radial velocity (upper panels). Thestars are then shown as a function of radius from the centers ofthe dSphs, with half-light radius drawn as a dashed line (mid-dle panels). Photometric [Fe/H] as described in the text is shownon the bottom panels, revealing the tight range in metallicitiesof And XV ([Fe/H] ∼ -1.6), And XVI ([Fe/H] ∼ -2.2). Symbols arehighlighted as in Figs 1&2.c (cid:13) , 000–000 Bruno Letarte et al.
Figure 4.
Spectra of candidate member stars in And XV. Toppanel shows robust members with cross correlation peak > . .Bottom panel shows more tentative member stars with cross cor-relation peak < . , but still lying on the well defined RGBof And XV. The inverse variance weighted summed spectrum isshown in the bottom offset panels for each subsample. Figure 5.
Spectra of candidate member stars in And XVI. Theinverse variance weighted summed spectrum is shown in the bot-tom offset panel.
Figure 6.
Likelihood distributions of member stars in And XV(top) and And XVI (bottom). The point shows the most likely val-ues of radial velocity and velocity dispersion, dashed lines showingthe 1, 2, 3 × σ distributions. For And XV, we show likelihood distri-butions of the seven best candidate members. The And XV disper-sion is resolved with v r =-339 +7 − km s − and σ v = 11 +7 − km s − .The And XVI dispersion is not quite resolved at σ with v r =-385 +5 − km s − and σ = 0 +10 − indef km s − .c (cid:13)000