1: \documentstyle[here,epsfig]{mn2e}
2: %\documentstyle[here,epsfig,referee]{mn2e}
3: \def\simlt{\mathrel{\rlap{\lower 3pt\hbox{$\sim$}}\raise 2.0pt\hbox{$<$}}}
4: \def\simgt{\mathrel{\rlap{\lower 3pt\hbox{$\sim$}} \raise 2.0pt\hbox{$>$}}}
5: \def\di{\mbox{d}}
6: \def\Msun{M_{\odot}}
7: \def\msun{M_{\odot}}
8: \def\HI{\hbox{H$\scriptstyle\rm I\ $}}
9: \def\gtsima{$\; \buildrel > \over \sim \;$}\def\gtsima{$\; \buildrel > \over
10: \sim \;$}
11:
12: \def\ltsima{$\; \buildrel < \over \sim \;$}
13: \def\gtrsim{\lower.5ex\hbox{\gtsima}}
14: \def\lesssim{\lower.5ex\hbox{\ltsima}}
15:
16: \newcommand{\q}{\begin{equation}}
17: \newcommand{\qa}{\begin{eqnarray}}
18: \newcommand{\qs}{\begin{eqnarray*}}
19: \newcommand{\nq}{\end{equation}}
20: \newcommand{\nqa}{\end{eqnarray}}
21: \newcommand{\nqs}{\end{eqnarray*}}
22: \newcommand{\ud}{\mathrm{d}}
23: \begin{document}
24:
25: \title[BSSs in dSphs]
26: {Blue straggler stars in dwarf spheroidal galaxies}
27:
28:
29: \author[M. Mapelli et al.]{M. Mapelli$^{1}$, E. Ripamonti$^{2}$, E. Tolstoy$^{2}$, S. Sigurdsson$^{3}$, M. J. Irwin$^{4}$, G. Battaglia$^{2}$
30: \\
31: $^1$Institute for Theoretical Physics, University of Z\"urich, Winterthurerstrasse 190, CH--8057 Z\"urich, Switzerland; {\tt
32: mapelli@physik.unizh.ch}\\ $^2$Kapteyn Astronomical Institute, University of Groningen, Postbus 800, 9700 AV Groningen, the Netherlands\\ $^3$Department of Astronomy and Astrophysics, The
33: Pennsylvania State University, 525 Davey Lab, University Park,
34: PA~16802, US\\$^{4}$Royal Greenwich Observatory, Madingley Road, Cambridge CB3 0EZ, UK\\}
35:
36: \maketitle
37: \vspace {7cm}
38:
39: \begin{abstract}
40:
41: Blue straggler star (BSS) candidates have been observed in all old dwarf
42: spheroidal galaxies (dSphs), however whether or not they are authentic
43: BSSs or young stars has been a point of debate. To both address this issue and
44: obtain a better understanding of the formation of BSSs in different
45: environments we have analysed a sample of BSS candidates in two nearby
46: Galactic dSphs, Draco and Ursa Minor. We have determined their radial
47: and luminosity distributions from wide field multicolour imaging data
48: extending beyond the tidal radii of both galaxies.
49:
50: BSS candidates are uniformly distributed through the host galaxy,
51: whereas a young population is expected to show a more clumpy
52: distribution.
53: Furthermore, the observed radial distribution of BSSs,
54: normalized to both red giant branch (RGB) and horizontal branch (HB)
55: stars, is almost flat, with a slight decrease towards the centre.
56: Such a distribution is at odds with the predictions for a young stellar
57: population, which should be more concentrated. Instead, it is
58: consistent with model predictions for BSS formation by mass transfer in
59: binaries (MT-BSSs). Such
60: %The observed spatial and luminosity distributions of BSSs are consistent with model predictions for formation by mass transfer in binaries (MT-BSSs), but not for formation models assuming stellar collisions (COL-BSSs). Our
61: results, although not decisive, suggest that these candidates are indeed BSSs and that MT-BSSs form
62: in the same way in Draco and Ursa Minor as in globular clusters. This
63: favours the conclusion that Draco and Ursa Minor are truly `fossil'
64: galaxies, where star formation ceased completely more than 8 billion
65: years ago.
66:
67: \end{abstract}
68:
69: \begin{keywords}
70: blue stragglers - stellar dynamics - galaxies: dwarf - galaxies: individual: Draco - galaxies: individual: Ursa Minor
71: \end{keywords}
72:
73: \section{Introduction}
74:
75: Blue straggler stars (BSSs) are stars located above and blue-ward of the
76: main sequence (MS) turn-off
77: in a color-magnitude diagram (CMD). They were first
78: discovered in a globular cluster (M3, Sandage 1953), and are mainly
79: observed in star clusters (Fusi Pecci et al. 1992; Ferraro et al. 1993, 1997; Zaggia, Piotto \& Capaccioli 1997; Ferraro et al. 2003, 2004; Sabbi et al. 2004; Hurley et al. 2005; Lanzoni et al. 2007a, 2007b and references therein), where the tiny (if any) spread
80: in the stellar age makes their identification straight forward.
81: However, there have been attempts to find halo BSSs in the Milky Way,
82: which show up as high-velocity stars brighter and hotter than turnoff
83: stars in the Galactic halo (Carney et al. 2001; Carney, Latham \& Laird
84: 2005; Beers et al. 2007).
85:
86:
87:
88:
89: Dwarf spheroidal galaxies (dSphs) seem
90: natural places to search for BSSs. Mateo et al. (1991) and Mateo,
91: Fischer \& Krzeminski (1995) first indicated the existence of a large
92: number of stars brighter than the turn-off mass in the Sextans
93: dSph. Mateo et al. (1995) suggested that these stars might be ordinary
94: MS stars substantially younger than the bulk of
95: the other stars.
96: %THIS SENTENCE MAKES YOU WONDER WHY YOU BOTHERED WITH THIS STUDY?
97: %However, recent studies tend to discard this interpretation
98: %identifying most of these stars with BSS (Lee et al. 2003
99: %and references therein; hereafter L03).
100: BSS candidates have been found in varying numbers in most dSphs, such as
101: Sculptor (e.g., Hurley-Keller, Mateo \& Grebel 1999; Monkiewicz et
102: al. 1999), Draco (Aparicio, Carrera \& Mart\'inez-Delgado 2001;
103: hereafter A01) and Ursa Minor (Carrera et al. 2002; hereafter C02).
104:
105: The issue of whether these stars are young or BSSs has not been
106: quantitatively addressed because there were not suitable models of
107: BSS formation with predictive power with which to compare the observations.
108: This also means that our proper understanding of
109: the star formation history of these `predominantly old' dSphs remains
110: in doubt. Has there actually
111: been low-level star formation in the last
112: 8$-$10 Gyr in these galaxies or can they really be considered `fossil'
113: galaxies?
114: Thus, in order to reconstruct the star formation history of
115: dSphs, it is crucial to understand whether the observed BSS candidates
116: are true BSSs rather than younger stars.
117:
118: A second unsolved question about BSSs is their formation mechanism
119: itself. BSSs are believed to have been somehow refuelled with hydrogen
120: after the MS phase. However, the refuelling mechanism is still
121: unknown. It has been proposed (McCrea 1964) that mass transfer in
122: isolated binaries can lead to the formation of BSSs. On the other hand,
123: BSSs could be the products of stellar collisions, occurring during (or
124: triggered by) three- and four-body encounters (Davies, Benz \& Hills
125: 1994; Sigurdsson, Davies \& Bolte 1994; Lombardi et al. 2002). In the
126: first case, i.e. mass-transfer BSSs (MT-BSSs), BSSs can form if
127: binaries are allowed to quietly evolve until they start the mass-transfer
128: phase. This implies that the local density should not be too
129: high, otherwise gravitational interactions will perturb the mass
130: transfer. In the second case, i.e. collisionally born BSSs (COL-BSSs),
131: the density must be sufficiently high to guarantee a short collision
132: time-scale.
133:
134: In globular clusters both processes might occur, as in the
135: high-density core we find the perfect conditions for COL-BSS formation,
136: while in the periphery MT-BSSs can originate from isolated binary
137: evolution. In some circumstances, the features of the observed BSS
138: population can be explained only by invoking a joint contribution by
139: these mechanisms (Leonard 1989; Fusi Pecci et al. 1992;
140: Bailyn \& Pinsonneault 1995; Ferraro et al. 1997; Sills \& Bailyn 1999;
141: Hurley et al. 2001). For example, in some globular clusters the
142: BSS radial distribution is bimodal (Ferraro et al. 1993, 1997;
143: Zaggia et al. 1997; Ferraro et al. 2004; Sabbi et al. 2004), with a central peak, a minimum at intermediate radii, and a further rise at the
144: periphery. Dynamical simulations by Mapelli et al. (2004, 2006, hereafter M04, M06, respectively) showed that this bimodal distribution can be
145: reproduced only by requiring the central BSSs to be
146: mainly COL-BSSs, and the peripheral BSSs to be
147: MT-BSSs. Also the luminosity function of BSSs in some globular clusters
148: (Bailyn \& Pinsonneault 1995; Sills \& Bailyn 1999; Sills et al. 2000;
149: Ferraro et al. 2003; Monkman et al. 2006) suggests the coexistence of
150: COL-BSSs and MT-BSSs.
151: %?? (generally fainter than the previous ones).
152:
153: In dSphs the collisional formation of BSSs should be impossible, as
154: the central density of these galaxies never reaches sufficiently high
155: values to allow stellar encounters. Thus, we expect BSSs in dSphs to be
156: solely MT-BSS type.
157:
158:
159: In this paper we seek to quantify the potential BSS population characteristics
160: in dSphs and thereby learn more about BSS formation and evolution, and
161: also about the star formation history in dSphs.
162: First of all, we check whether the main
163: characteristics (such as radial and luminosity distribution) of BSS
164: candidates in dSphs are more consistent with those predicted by theoretical
165: models of BSSs, or if they are more similar to MS
166: young stars. At the same time, we would like to test whether the
167: proposed formation mechanisms for BSSs also work in dSphs,
168: i.e. whether the BSS candidates in dwarf galaxies can be
169: connected with MT-BSSs.
170:
171: We focus on the BSS population of two dSphs, Ursa Minor and Draco (see Table~1).
172: These are among the faintest and most diffuse dwarfs in the Local Group
173: (Irwin \& Hatzidimitriou 1995, hereafter IH95; Mateo 1998). They also appear to be among the most dark matter dominated objects we
174: know about (Kleyna et al. 2003; Wilkinson et al. 2004). Ursa Minor and
175: Draco are also among those dSphs of the Local Group where star formation
176: %has been the least efficient and - not really !
177: appeared to cease early on ($\gtrsim{}8-10$ Gyr
178: ago, see Mateo 1998; Hernandez, Gilmore \& Valls-Gabaud 2000; A01; C02;
179: Bellazzini et al. 2002). In both these galaxies a significant number of
180: BSS candidates have been detected (A01; C02). If the observed BSS
181: candidates in these two dSphs can be explained as young MS stars,
182: the existence of BSSs in dwarf galaxies can probably be safely rejected.
183: However, if instead these stars do behave like authentic BSSs, then Ursa Minor
184: and Draco should be really considered two 'fossil' galaxies, where star
185: formation indeed completely stopped many Gyr ago. Furthermore, by studying
186: such diffuse systems we can also learn about the
187: properties of BSSs in a much less dense environment than a globular
188: cluster.
189:
190: %Given the poor star formation history of Ursa Minor and Draco, if the BSS candidates in Ursa Minor and Draco were shown to be cannot be e
191:
192:
193: In Section 2 we present the data on which our analysis is based
194: . In Section 3 we discuss the observational features of
195: BSS candidates, with particular emphasis on the luminosity function and radial
196: distribution characteristics. In Section 4 we
197: describe our dynamical simulations and compare the results with
198: observations. A comparison with previous work on BSS
199: candidates in dwarf galaxies and globular clusters, is presented in
200: Section 5. Finally in Section 6 we present our conclusions.
201:
202: \section{The data}
203: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% FIGURE 1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
204: \begin{figure*}
205: \center{{
206: \epsfig{figure=fig1a.eps,height=8cm}
207: \epsfig{figure=fig1b.eps,height=8cm}
208: \epsfig{figure=fig1c.eps,height=8cm}
209: \epsfig{figure=fig1d.eps,height=8cm}
210: }}
211: \caption{\label{fig:spatial} Upper panel: Right Ascension and
212: Declination of the stars imaged in Draco (left) and Ursa Minor
213: %these are presumably \deltaRA *cos(dec) and \deltaDec
214: (right). The concentric ellipses indicate tidal and core radii ($r_t$
215: % you can't see the core ellipse for Draco
216: and $r_c$; the adopted values are listed in Table 1).
217: %; for Draco $r_c=7.63$ arcmin, $r_t=45.1$ arcmin (S\'egall et al. 2007) and for Ursa Minor $r_c=15.8$ arcmin, $r_t=50.6$ arcmin (IH95).
218: In both cases the origin of the axes coincides with the centre
219: of the observed galaxy. Bottom panel: CMDs of Draco (left) and Ursa Minor
220: (right).}
221: \end{figure*}
222: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
223: \subsection{INT/WFC survey data}
224: %I.e., where do they come from, published/unpublished, errors etc. etc.
225: The Isaac Newton Telescope (INT) Wide Field Camera (WFC) is a mosaic of four 4k $\times$ 2k EEV chips,
226: offering a field of view of $\sim$0.29 square degrees. It is mounted in the
227: prime focus of the 2.5-m Isaac Newton Telescope on La Palma. The $V$'-band
228: (Harris filter) and $i$'-band (SDSS-like)\footnote{For filter responses see
229: {\tt http://www.ast.cam.ac.uk/$\sim{}$wfcsur/technical/filters/}} data extend
230: beyond the tidal radius in both Draco and Ursa Minor
231: (see the upper panels of Fig.~\ref{fig:spatial}). They were taken
232: as part of the INT Wide Field Survey (McMahon et al 2001) during 2002 with
233: an average seeing of 1~arcsec and in generally photometric conditions.
234: The images were processed in the standard way with the INT WFC pipeline
235: (Irwin \& Lewis 2001). The two-dimensional instrumental signature
236: removal includes provision for: non-linearity correction at the detector
237: level; bias and overscan correction prior to trimming to the active
238: detector areas; flatfielding; and fringe removal in the $i$'-band.
239:
240: Catalogue generation follows the precepts outlined by Irwin (1985, 1996)
241: and includes the facility to: automatically track any background variations
242: on scales of typically ~20-30 arcsec; detect and deblend images or groups
243: of images; and parameterise the detected images to give various (soft-edged)
244: aperture fluxes, position and shape measures. The generated catalogues
245: start with an approximate World Coordinate System (WCS) defined by the
246: known telescope and camera properties (e.g. WCS distortion model) and are
247: then progressively refined using all-sky astrometric catalogues [e.g. United States Naval Observatory (USNO) catalogue of astrometric standards,
248: Automated Plate Measuring Machine (APM) catalogue, Two Micron All Sky Survey (2MASS)] to give internal precision generally better than 0.1 arcsec and
249: global external precision of ~0.25 arcsec or better. These latter numbers
250: are solely dependent on the accuracy of the astrometric catalogues used in
251: the refinement.
252:
253: All catalogues for all CCDs for each pointing are then processed using
254: the image shape parameters for morphological classification in the main
255: categories: stellar; non-stellar; noise-like. A sampled curve-of-growth
256: for each detected object is derived from a series of aperture flux
257: measures as a function of radius. The classification is then based on
258: comparing the curve-of-growth of the flux for each detected object with
259: the well-defined curve-of-growth for the general stellar locus. This
260: latter is a direct measure of the integral of the point spread function
261: (PSF) out to various radii and is independent of magnitude, if the
262: data are properly linearised, and if saturated images are excluded. The
263: average stellar locus on each detector is clearly defined and is used as
264: the basis for a null hypothesis stellar test for use in classification.
265: The curve-of-growth for stellar images is also used to automatically
266: estimate frame-based aperture corrections for conversion to total flux.
267:
268: The photometric standards observed during the run (mainly Landolt 1992
269: and spectrophotometric standards) are automatically located in a standards
270: database and used to estimate the zero-point in each passband for every
271: pointing containing any of these standards. The trend in the derived
272: zero-points is then used to assign a photometric quality index for
273: each night and also as a first pass estimate for the magnitude calibration
274: for all the observations.
275:
276: Various quality control plots are generated by the pipeline and these
277: are used to monitor characteristics such as the seeing, the average
278: stellar image ellipticity (to measure trailing), the sky brightness
279: and sky noise, the size of aperture correction for use with the 'optimal'
280: aperture flux estimates (here 'optimal' refers to the well-known property
281: that soft-edged apertures of roughly the average seeing radius provide
282: close to profile fit accuracy, e.g. Naylor 1998).
283:
284: The `optimal' catalogue fluxes for the $V$', $i$' filters for each
285: field are then combined to produce a single matched catalogue
286: for each pointing and the overlaps between pointings are used to
287: cross-calibrate all the catalogues to a common system with typical accuracy
288: 1-2 per cent across the survey region. The final step is to produce a unique
289: catalogue for the whole region by removing spatially coincident (within 1
290: arcsec) duplicates.
291:
292: As a final stage the data are converted\footnote{For details of the conversion
293: see {\tt http://www.ast.cam.ac.uk/$\sim{}$wfcsur/technical/photom/}} from the instrumental
294: WFC $V$' and $i$' passbands to the Johnson-Cousins $V$,$I$ system to obtain the
295: standard CMD shown in the lower panel of
296: Fig.~\ref{fig:spatial}.
297:
298: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% FIGURE 2 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
299: \begin{figure*}
300: \center{{
301: \epsfig{figure=fig2a.eps,height=8cm}
302: \epsfig{figure=fig2b.eps,height=8cm}
303: }}
304: \caption{\label{fig:fig1} CMD of Draco (left panel) and Ursa Minor
305: (right) with stellar population selection boxes overlaid. Boxes indicated
306: by the solid line and labelled as 1 and 2 are
307: the BSSs and RGB stars, respectively. Boxes indicated by the dashed line
308: and labelled as 3 are the HB stars, divided as blue and red.}
309: \end{figure*}
310: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
311:
312: \subsection{Stellar population selection criteria}
313:
314: From these data
315: %in addition to samples of {\bf contaminating} foreground
316: %stars\footnote{{\bf Our data are also contaminated both by Milky Way
317: %stars in the foreground, and by extragalactic objects (e.g. quasars and
318: %unresolved galaxies) in the background. Since the foreground component
319: %is dominant, in the rest of this paper we will refer to any type of
320: %contamination as ``foreground'', unless the distinction is important.}}
321: we selected three different populations: BSSs, red giant branch (RGB)
322: and horizontal branch (HB) stars. The last two populations are
323: considered good tracers of the overall light from the galaxy, and we use them as a
324: comparison for BSS distributions. For all these populations we adopt
325: more conservative selection criteria with respect to previous works
326: (see e.g. A01; C02; Lee et al. 2003, hereafter L03), in order to
327: minimize the contamination by stars of different populations. The
328: regions of the CMD we associate with BSSs, RGB and HB stars are
329: indicated in Fig.~\ref{fig:fig1} as boxes 1, 2 and 3, respectively. In
330: particular, for BSSs we chose the $V$ and ($V$-$I$) range to be (i)
331: sufficiently above the turn-off, in order to avoid contamination from
332: the MS, (ii) blue-ward of the RGB, avoiding not only contamination from
333: these stars but especially the region of the CMD most affected by the
334: Galactic foreground, (iii) red-ward of a possible faint extension of the
335: very blue extended HB.
336:
337: Furthermore, we select a narrow strip of RGB stars (box 2), to limit the
338: effect of binaries and errors in magnitude. The large number of RGB
339: stars in the sample allows us this conservative choice. Finally, the HB
340: region is divided in two different regions, approximately corresponding
341: to the red HB (RHB) and the blue HB (BHB). As it has already been noted
342: (C02), the HB in Ursa Minor is substantially bluer than in Draco.
343:
344: We also checked that these more restrictive selection criteria do not
345: significantly affect our results both for the radial and for the
346: luminosity distribution (see next section for the comparison with L03).
347:
348: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% TABLE 1%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
349: \begin{table*}
350: \begin{center}
351: \caption{Galaxy parameters} \leavevmode
352: \begin{tabular}[!h]{lllllllll}
353: \hline
354: Galaxy
355: & $d$$^{\rm a}$ (kpc)
356: & $r_{c}$$^{\rm b}$ (arcsec)
357: & $r_{t}$$^{\rm b}$ (arcsec)
358: & $\sigma_c$ (km s$^{-1}$)$^{\rm c}$
359: & $n_{c}$ (stars pc$^{-3}$)$^{\rm d}$
360: & $W_0$$^{\rm e}$
361: & $c$$^{\rm e}$
362: & ellipticity$^{\rm b}$ \\
363: \hline
364: Draco & 83 & 457.8 & 2706 & 10.5 & $3.2\times{}10^{-3}$ & 2.0 & 0.76 & 0.29 \\
365: Ursa Minor & 76 & 948 & 3036 & 12.5 & $10^{-3}$ & 0.45 & 0.52 & 0.56 \\
366: \noalign{\vspace{0.1cm}}
367: \hline
368: \end{tabular}
369: \end{center}
370: \footnotesize{ $^{\rm a}$ We assume distance moduli of 19.60 (Draco) and
371: 19.41 (Ursa Minor); see appendix B; $^{\rm
372: b}$Core radius ($r_c$), tidal radius ($r_t$) and ellipticity are from
373: S\'egall et al. (2007) and from IH95 for Draco and Ursa Minor,
374: respectively. $^{\rm c}$Core velocity dispersion of the dSph, from
375: Wilkinson et al. (2004). $^{\rm d}$Core density (n$_c$)of the dSph,
376: derived from our data. $^{\rm e}$Central adimensional potential ($W_0$)
377: and concentration [$c={\rm log}(r_c/r_t)$] are derived from our simulations. $c$ is
378: consistent with IH95 for Ursa Minor and with
379: S\'egall et al. (2007) for Draco. }
380: \end{table*}
381: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
382: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% FIGURE 2 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
383: %\begin{figure*}
384: %\center{{
385: %%\epsfig{figure=draco_surfdensity.ps,height=8cm}
386: %%\epsfig{figure=ursaminor_surfdensity.ps,height=8cm}
387: %\epsfig{figure=radial_distr_draco.eps,height=8cm}
388: %\epsfig{figure=ursaminor_surfdensity.ps,height=8cm}
389: %}}
390: %\caption{\label{fig:fig2} Surface density of BSS (dotted line connecting circles), HB
391: %(dashed line connecting squares) and RGB stars (solid line connecting triangles) in Draco (left panel) and Ursa Minor
392: %(right). Size of points and poissonian errors are comparable.
393: %}
394: %\end{figure*}
395: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
396: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% FIGURE 3 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
397: \begin{figure*}
398: \center{{
399: \epsfig{figure=fig3a.eps,height=8cm}
400: \epsfig{figure=fig3b.eps,height=8cm}
401: }}
402: \caption{\label{fig:fig3}
403: Observed relative frequency of BSSs normalized to RGB stars (filled squares connected by dashed line) and HB stars (filled triangles connected by dotted line). Observed relative frequency of HB stars normalized to RGB stars (filled circles connected by solid line). Left panel refers to Draco, right panel to Ursa Minor.
404: }
405: \end{figure*}
406: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
407:
408:
409: \section{Observational properties of BSS candidates}
410:
411: After accounting for the foreground and background contamination\footnote{Our selection boxes are contaminated both by Milky Way stars in the
412: foreground, and by extragalactic objects (e.g. quasars and unresolved
413: galaxies) in the background. Since the foreground component is generally
414: dominant,
415: in the rest of this paper we will refer to any type of contamination as
416: `foreground', unless the distinction is important.} in our data,
417: (see Appendix A for the details), we can extract information about the
418: radial\footnote{ All the references to `radii' in this paper mean {\it
419: elliptical radii}. The elliptical radius of a point $(x,y)$ is
420: $r_{ell}(x,y)^2 = x^2 + [y/(1-e)]^2$, where $e$ is the ellipticity of
421: the considered galaxy, and the galaxy is assumed to be centred on the
422: origin, with its major axis aligned with the $x$-axis.} distribution of
423: different populations of stars as well as about their luminosity
424: distribution. Both these quantities are important to understand the
425: behaviour of BSS candidates [see e.g. M06 for the radial distribution
426: and Monkman et al. (2006) for the luminosity].
427:
428: \subsection{Radial distributions}
429: %%Fig.~\ref{fig:fig2} shows the surface density profile of BSS,
430: %%RGB and HB stars in Draco (left panel) and Ursa Minor (right panel). The
431: %quantities used in Fig.~2
432: %%(in particular the central
433: %%radius of each annulus and the number of BSS, RGBs and HBs per annulus)
434: %are listed in Table~2 and Table~3 for Draco and Ursa Minor,
435: %respectively.
436: Fig.~\ref{fig:fig3} shows the radial distribution of the ratio between
437: the number of BSSs ($N_{{\rm BSS}}$) and that of RGB ($N_{{\rm RGB}}$)
438: and of HB stars ($N_{{\rm HB}}$). The radial distribution of $N_{{\rm
439: HB}}$ with respect to $N_{{\rm RGB}}$ is also shown in
440: Fig.~\ref{fig:fig3}. The quantities used in Fig.~\ref{fig:fig3} are
441: listed in Table~2 and Table~3 for Draco and Ursa Minor respectively.
442:
443: The behaviour of these three relative frequencies
444: is qualitatively similar in Ursa Minor (right panel) and Draco (left
445: panel).
446: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% TABLE 2%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
447: \begin{table*}
448: \begin{center}
449: \caption{Number counts for Draco.}
450: \leavevmode
451: \begin{tabular}[!h]{lllllll}
452: \hline
453: $r\,{}$(arcsec)$^{\rm a}$
454: & $N_{{\rm BSS}}$$^{\rm b}$
455: & $\epsilon_{{\rm BSS}}$$^{\rm c}$
456: & $N_{{\rm RGB}}$$^{\rm b}$
457: & $\epsilon_{{\rm RGB}}$$^{\rm c}$
458: & $N_{{\rm HB}}$$^{\rm b}$
459: & $\epsilon_{{\rm HB}}$$^{\rm c}$\\
460: \hline
461: 198 & 30.8 (31) & 5.6 & 316 (319) & 17.9 & 152 (155) & 12.5\\
462: 522 & 33.5 (34) & 5.8 & 222 (227) & 15.1 & 128 (134) & 11.7\\
463: 800 & 25.1 (26) & 5.1 & 130 (139) & 11.8 & 80.3 (94) & 10.1\\
464: 1080 & 15.1 (16) & 4.0 & 65.3 (75) & 8.7 & 48.0 (56) & 7.7\\
465: 1450 & 11.5 (14) & 3.8 & 67.2 (93) & 9.9 & 49.3 (72) & 9.2\\
466: \noalign{\vspace{0.1cm}}
467: \hline
468: \end{tabular}
469: \end{center}
470: \footnotesize{
471: $^{\rm {a}}$Centre of the annulus. $^{\rm {b}}$The value out of (in) the parenthesis is after (before) the subtraction of the foreground. $^{\rm {c}}$ Poissonian error plus a term accounting for foreground subtraction.
472: }
473: \end{table*}
474: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
475:
476: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% TABLE 3%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
477: \begin{table*}
478: \begin{center}
479: \caption{Number counts for Ursa Minor.}
480: \leavevmode
481: \begin{tabular}[!h]{lllllll}
482: \hline
483: $r\,{}$(arcsec)$^{\rm a}$
484: & $N_{{\rm BSS}}$$^{\rm b}$
485: & $\epsilon_{{\rm BSS}}$$^{\rm c}$
486: & $N_{{\rm RGB}}$$^{\rm b}$
487: & $\epsilon_{{\rm RGB}}$$^{\rm c}$
488: & $N_{{\rm HB}}$$^{\rm b}$
489: & $\epsilon_{{\rm HB}}$$^{\rm c}$\\
490: \hline
491: 225 & 17.9 (18) & 4.2 & 182 (183) & 13.5 & 63.5 (64) & 8.0\\
492: 650 & 29.8 (30) & 5.5 & 237 (239) & 15.5 & 106.0 (110) & 10.6\\
493: 1020 & 23.7 (24) & 4.9 & 146 (149) & 12.2 & 79.5 (84) & 9.3\\
494: 1350 & 14.7 (15) & 3.9 & 82.8 (86) & 9.3 & 40.0 (41) & 6.4\\
495: 1700 & 9.7 (10) & 3.2 & 40.8 (44) & 6.6 & 27.0 (29) & 5.5\\
496: 2370 & 2.8 (3) & 1.7 & 19.1 (21) & 4.6 & 13.0 (15) & 4.0\\
497:
498: \noalign{\vspace{0.1cm}}
499: \hline
500: \end{tabular}
501: \end{center}
502: \footnotesize{
503: $^{\rm {a}}$Centre of the annulus. $^{\rm {b}}$The value out of (in) the parenthesis is after (before) the subtraction of the foreground. $^{\rm {c}}$ Poissonian error plus a term accounting for foreground subtraction.
504: }
505: \end{table*}
506: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
507:
508: From the shape of the distributions in Fig.~\ref{fig:fig3} we can see that the
509: BSS candidates
510: appear to be slightly less concentrated than both HB and RGB stars. The
511: relative frequency of BSSs is low especially within 1 $r_c$,
512: and there are hints of a maximum at a distance 1.5
513: $r_c\lesssim{}r\lesssim{}\,{} 2.5\,{}r_c$. The most remarkable
514: feature of this distribution is the absence of a central peak in the
515: relative BSS frequency, which is present in nearly every
516: globular cluster.
517: %The absence of the central peak is crucial to
518: %understand the nature of these stars.
519:
520: The distributions of $N_{\rm BSS}/N_{\rm HB}$ and $N_{\rm BSS}/N_{\rm RGB}$ are marginally consistent with flat distributions. In fact, if we try to fit $N_{\rm BSS}/N_{\rm RGB}$ with a flat distribution, the minimum non-reduced $\chi{}^2$ is 8.4 (corresponding to a level of the flat distribution equal to 0.129) and 4.9 (for a level of the flat distribution equal to 0.130), for Draco and Ursa Minor\footnote{The data points used in the $\chi{}^2$ analysis are 5 for Draco and 6 for Ursa Minor (see Fig.~\ref{fig:fig3}). There is 1 parameter, i.e. the level of the flat distribution.}, respectively.
521: %According to the Kolmogorov-Smirnov (KS) test, the probability that $N_{\rm BSS}/N_{\rm RGB}$ is drawn from a flat distribution is $\sim{}0.7$ for Draco and only $\sim{}0.3$ for Ursa Minor.
522: The resultant null hypothesis probability for a flat distribution is only $\sim{}0.08$ for Draco and $\sim{}0.43$ for Ursa Minor.
523:
524: % I would have said a better way to say this is to argue that these results
525: % imply you cannot reject the flat distribution hypothesis. I also would
526: % include the number of degrees of freedom in the text not a footnote and
527: % quote the reduced chisqu since its easier to interpret. Why not also
528: % quote the KS results in the text because they support even more the fact
529: % that a flat distribution fits the data ?
530:
531: Similarly, to fit $N_{\rm BSS}/N_{\rm HB}$ with a flat distribution, the minimum non-reduced $\chi{}^2$ is 2.6 (corresponding to a level of the flat distribution equal to 0.243) and 1.0 (for a level of the flat distribution equal to 0.288), for Draco and Ursa Minor, respectively. The resultant null hypothesis probability for a flat distribution is $\sim{}0.63$ for Draco and $\sim{}0.96$ for Ursa Minor.
532: %With the KS test, the probability that $N_{\rm BSS}/N_{\rm HB}$ is drawn from a flat distribution is only $\sim{}0.04$ for Draco and only $\sim{}0.3$ for Ursa Minor.
533:
534: %We also checked what is the maximum value of the central bin of $N_{\rm BSS}/N_{\rm RGB}$ ($N_{\rm BSS}/N_{\rm HB}$)
535:
536:
537:
538: We also checked the probability that $N_{\rm BSS}/N_{\rm RGB}$ and $N_{\rm BSS}/N_{\rm HB}$ are consistent with a distribution rising in the central bin and flat elsewhere\footnote{For this analysis we have 2 parameters: the level of the central bin and the level of the flat distribution for the other bins.}. For Draco, we found that this probability drops below 0.05 if the central bin is a factor of 1.00 (1.49) higher than the outer ones in the case of $N_{\rm BSS}/N_{\rm RGB}$ ($N_{\rm BSS}/N_{\rm HB}$). For Ursa Minor, the probability drops below 0.05 if the central bin is a factor of 1.45 (2.42) higher than the outer ones in the case of $N_{\rm BSS}/N_{\rm RGB}$ ($N_{\rm BSS}/N_{\rm HB}$). Then, we can conclude that the observed distribution of $N_{\rm BSS}/N_{\rm HB}$ and especially $N_{\rm BSS}/N_{\rm RGB}$ are hardly consistent with a central rise like the one we observe in most of globular clusters (M06).
539:
540:
541: If BSS candidates were young MS stars, we would expect them to be more
542: concentrated than older stars,
543: %because the gas is likely to be exhausted later in the centre than
544: %in the outskirts, and also
545: consistent with observations where
546: metal-rich (younger) stars
547: are typically more centrally concentrated than metal-poor (older) stars
548: (e.g. Tolstoy et al. 2004, Battaglia et al. 2006).
549:
550:
551:
552: Furthermore, the spatial distribution of
553: BSS does not show the clumping which could be expected in the case of a
554: young population (e.g., Fornax dSph; Battaglia et al. 2006). Thus, the
555: observed radial distribution of BSS candidates suggests (even if does not prove) that these stars are
556: BSSs and not a young population.
557: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% FIGURE 4 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
558: \begin{figure*}
559: \center{{ \epsfig{figure=fig4a.eps,height=8cm}
560: \epsfig{figure=fig4b.eps,height=8cm} }}
561: \caption{\label{fig:fig5} Luminosity distribution of BSSs in Draco (left
562: panel) and Ursa Minor (right panel). The empty histogram represents the
563: entire sample of BSSs and the error bars show the Poissonian
564: error. The lightly hatched (heavily hatched) histogram represents BSSs
565: with radial position $r>r_c$ ($r<r_c$). }
566: \end{figure*}
567: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
568:
569: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% FIGURE 5 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
570: \begin{figure*}
571: \center{{
572: \epsfig{figure=fig5a.eps,height=8cm}
573: \epsfig{figure=fig5b.eps,height=8cm}
574: }}
575: \caption{\label{fig:fig6} Surface density profile of Draco (left panel)
576: and Ursa Minor (right panel). The number density is given in stars per
577: arcmin$^2$. The filled circles are data points from IH95. The dashed
578: line is the best-fitting simulation. }
579: \end{figure*}
580: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
581:
582:
583: %Assuming that these stars are BSSs,
584: Are these observed radial distributions
585: consistent with models of BSS formation and evolution? According to the
586: model by M04 and M06, most BSSs in the core of a dense
587: host system form from collisions. A central peak in the relative BSS
588: frequency is expected only if a sufficiently large number of COL-BSSs can
589: form.
590:
591:
592: The absence of any central peak in Fig.~\ref{fig:fig3} is consistent with
593: this model. In fact, in dSphs the stellar density, even in the
594: core, is so low that stellar collisions are unlikely to occur\footnote{The collision rate, defined as (Davies, Piotto \& De Angeli 2004) $\Gamma{}\sim{}N_c\,{}n_c\,{}\Sigma{}_{coll}\,{}\sigma{}_c$ (where $\Sigma{}_{coll}$ is the collision cross-section, $N_c$ the number of stars in the core, $n_c$ and $\sigma{}_c$ the stellar number density and velocity dispersion in the core), is more than a factor of $10^5$ smaller in dSphs than in globular clusters.}, and
595: COL-BSSs cannot form. The
596: dearth of BSSs in the centre is also favoured by
597: the inefficiency of dynamical friction in dSphs: even if BSSs have higher
598: mass than both RGB and HB stars, it takes too long for them to
599: sink to the centre.
600:
601: Furthermore, M04 and M06 also predict that MT-BSSs (see Section 4) have a
602: relative frequency that is almost flat in the centre and slightly rising
603: in the periphery. This implies that BSS candidates in Draco and Ursa
604: Minor do behave like MT-BSSs, whereas they are unlikely to be COL-BSSs.
605:
606:
607: %The BSS candidates in Draco and Ursa Minor are observed to have a relative
608: %frequency that is almost flat in the centre and slightly rising in the
609: %periphery, which is consistent with models for MT-BSS (see Section
610: %4, and M06).
611:
612: %The presence of such peak within the core radius is considered
613: %associated with COL-BSS, which form only in the densest region of the
614: %host system (M04, M06). Thus, the absence of the central peak seems to exclude the existence of COL-BSS in these dwarf galaxies, as we would have expected, given the low density of these system.
615: %{\it Vice versa} the absence of any central peak in such a low density environment supports the theory that most of BSS in the core of the host system form from collisions. The lack of BSS in the centre is also encouraged by the inefficiency of dynamical friction in dSphs: even if BSS have higher mass than both RGBs and HBs, it takes too much time for them to sink to the centre.
616:
617:
618: %To check if this consideration holds also for Draco and Ursa Minor, we divided our sample of RGBs (HBs) in blue RGBs (BHBs) and in (RHBs), and we calculated the radial distribution..E' IL CASO DI FARLO?
619:
620: %L03 also claims that the radial distribution of BSS inside the
621: %Sextans dSph depends on their luminosity, i.e. that faint BSS are less
622: %concentrated than the bright ones. We have checked whether this is true
623: %also in the case of Draco and Ursa Minor.
624:
625: We compared our definitions of BSSs and other populations with those
626: used by L03 and tested the robustness of different choices. L03 adopt
627: a wider definition of BSS, and normalize the frequency of BSSs to the
628: number of sub-giant branch (SGB) stars. We used both our definition of
629: BSSs and normalization to the RGB stars and the L03 definition
630: adapted for the distances and reddenings of Draco/Ursa Minor, combined
631: with a normalization to SGB stars.
632: %In Fig.~\ref{fig:fig4} we report the results
633: %for Draco (left) and Ursa Minor (right panel), adopting the L03
634: %definition and normalization (but the results do not change if we adopt
635: %our definitions).
636: We do not observe any significant difference.
637: % between faint ($22.1<V<22.8$) and bright ($20.9<V<22.1$)
638: %BSS. Indeed, bright BSS in Draco seem slightly less concentrated than
639: %the faint ones. The discrepancy between our results and that of L03
640: %could be explained with an intrinsic difference of BSS in
641: %Sextans with respect to Draco and Ursa Minor (see Section 5).
642:
643: %The distribution of $N_{{\rm BSS}}/N_{{\rm SGB}}$
644: %%($N_{{\rm SGB}}$ is the number of SGB stars per each annulus)
645: %is very
646: %similar to the distribution of both $N_{{\rm BSS}}/N_{{\rm RGB}}$ and
647: %$N_{{\rm BSS}}/N_{{\rm HB}}$. These properties suggest
648:
649: This suggests that our results
650: are reasonably independent of the BSS selection
651: criteria, and also of the stellar population we choose as a normalization
652: control sample (HB, RGB or SGB).
653:
654:
655: In Fig.~\ref{fig:fig3} it can also be seen that HB stars also seem to be
656: slightly less concentrated than RGB stars.
657: %, especially in Ursa Minor.
658: %The mass
659: %and the age of these two groups, in fact, should be similar,
660: %and the difference in the two spatial distributions is
661: %puzzling. However, we note that the error bars of $N_{{\rm HB}}/N_{{\rm
662: %RGB}}$ are quite large (especially in the case of Ursa
663: %Minor), so that the distribution is still consistent with a flat one.
664: This again is consistent with trends seen in other dSph
665: (Harbeck et al. 2002;
666: L03; Tolstoy et al. 2004; Battaglia et al. 2006), suggesting that older
667: populations tend to be less centrally concentrated.
668: In particular the blue old stellar populations (BHB, blue RGB, etc)
669: tend to be less concentrated than their red counterparts (RHB, red RGB, etc).
670: However, as in the case of $N_{\rm BSS}/N_{\rm HB}$ and $N_{\rm BSS}/N_{\rm RGB}$, $N_{\rm HB}/N_{\rm RGB}$ is also statistically consistent with a flat distribution.
671: %{\bf [ER: red/blue RGBs??? btw, il Tolstoy et al. 2004 ed il Battaglia et al. 2006 confermano la cosa non solo attraverso la fotometria ma anche dagli spettri]}
672:
673: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% FIGURE 4 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
674: %\begin{figure*}
675: %\center{{
676: %\epsfig{figure=lee.eps,height=8cm}
677: %\epsfig{figure=lee_ursa_nofg.eps,height=8cm}
678: %}}
679: %\caption{\label{fig:fig4}
680: %Observed relative frequency of BSS normalized to SGBs. Left panel refers to Draco, right panel to Ursa Minor. The solid line refers to all BSS, the dotted line to faint BSS and the dashed line to bright BSS. BSS are defined [accordingly to L03] as stars with $20.9<$V$<22.8$ and with $-0.1<$(V-I)$<0.2$ for Draco and $0.1<$(V-I)$<0.4$ for Ursa Minor. SGBs are RGBs with $20.9<$V$<22.8$. Faint (bright) BSS are BSS with $22.1<$V$<22.8$ ($20.9<$V$<22.1$).
681: %}
682: %\end{figure*}
683: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
684:
685: \subsection{Luminosity distribution}
686: The luminosity distribution is another important indicator of BSS
687: properties. Recent papers (Ferraro et al. 2003; Monkman et al. 2006)
688: suggest a correlation between the brightness of BSSs and their
689: radial distance from the the centre of a
690: globular cluster. Bright BSSs tend to be
691: more concentrated than the faint ones. In
692: turn, if the model of M04 and M06 is correct, such a correlation suggests
693: that the centrally concentrated COL-BSSs tend to be brighter than
694: MT-BSSs. This scenario makes sense, as
695: COL-BSSs should conserve a large fraction of the mass of the colliding
696: progenitors (Benz \& Hills 1987, 1992; Sills et al. 2001; Freitag \&
697: Benz 2005), whereas the mass-transfer process is not as efficient (Pols
698: \& Marinus 1994; Tian et al. 2006).
699:
700: In dSphs, where only MT-BSSs are likely to form, we do not expect to see
701: a correlation between the brightness of BSSs and their radial
702: position. This prediction is completely supported by the observed BSS
703: luminosity distribution of Draco (Fig.~\ref{fig:fig5}, left panel) and
704: Ursa Minor (Fig.~\ref{fig:fig5}, right panel). The open histograms in
705: Fig.~\ref{fig:fig5} show the total luminosity distribution, while the
706: light and heavy hatched histograms show the luminosity distribution of
707: BSSs which are located outside and within $r_c$, respectively. According
708: to the Kolmogorov-Smirnov (KS) test, the probability that light and heavy hatched histograms are drawn from the same distributions is $>0.999$
709: %9994 and 0.9999565
710: both for Draco and for Ursa Minor. Thus, there is
711: % couldn't we simply say < 1.0e-6 and < 1.0e-4 respectively
712: no statistically significant difference between these distributions,
713: indicating no dependence of the luminosity function on the radial
714: distance. This fact is at odds with the findings by L03, who observe a
715: correlation between the brightness of BSSs in Sextans and their radial
716: distance. Since (as we checked) our luminosity distributions do not
717: change by adopting the BSS selection criteria by L03, we suggest that
718: this is due to an intrinsic difference between Sextans and Draco/Ursa
719: Minor (see Section 5).
720:
721: %In order to be sure that our selection criteria do not affect significantly our results, we made also some checks by adopting different criteria. For example, for comparison with L03, we considered Draco BSS all those stars with V=[ 20.9, 22.8] and (V-I)=[-0.3, 0.3]
722:
723:
724: \section{The simulations}
725:
726: The data presented in the previous sections
727: show that BSS candidates in Draco and Ursa Minor behave like
728: MT-BSSs. As a further check, we ran for Draco and
729: Ursa Minor the same kind of dynamical simulations that were performed by
730: M04 and M06 for BSSs in globular clusters.
731:
732: \subsection{Method}
733: We adopt the upgraded version of the code by Sigurdsson \& Phinney
734: (1995) already described in M04 and M06. The code integrates the
735: dynamics of BSSs, under the influence of the galactic potential, of
736: dynamical friction (using Chandrasekhar formula) and of distant
737: encounters with other stars. Also three-body encounters are implemented in
738: the code; but they are unimportant in the runs for dSphs.
739: %{\bf We expect that for dSphs also dynamical friction and distant encounters are much less efficient than in globular clusters}.
740:
741:
742: The potential of the host galaxy is represented by a time independent
743: multimass King model. The classes of mass are the same as in M04, and
744: the assumed turn-off mass is 0.8 $M_\odot{}$. To calculate the
745: potential, we input the observed core density ($n_c$) and velocity
746: dispersion ($\sigma{}_c$) of Draco and Ursa Minor (the adopted values
747: are listed in Table~1), and we modify the value of the
748: central adimensional potential, $W_0$ (defined in Sigurdsson \& Phinney
749: 1995), until we reproduce the concentration and the
750: density profile of the galaxy under consideration.
751: In Fig.~\ref{fig:fig6}
752: the density profiles
753: of the best-fitting King models are compared with the data of
754: IH95. As expected, the best-fitting
755: value of $W_0$ is a factor 5$-$20 lower than the common values assumed in
756: globular clusters.
757:
758: BSSs are generated with a given position, velocity and mass. Initial
759: positions are randomly chosen according to a probability distribution
760: homogeneous in the radial distance from the centre. This means that BSSs
761: are initially distributed according to an isothermal sphere, as we
762: expect for MT-BSSs (see M04, M06). The minimum and the maximum value of
763: the distribution of initial radial distances, $r_{min}$ and $r_{max}$,
764: have been tuned in order to find the best-fitting simulation (Table 4
765: and 5 report the most significant runs and their parameters for Draco
766: and Ursa Minor, respectively).
767:
768: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% TABLE 4%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
769: \begin{table*}
770: \begin{center}
771: \caption{Simulation parameters and $\chi{}^2$ for Draco.}
772: \leavevmode
773: \begin{tabular}[!h]{llllllll}
774: \hline
775: Run
776: & $r_{min}/r_c$
777: & $r_{max}/r_c$
778: & $v_{kick}/\sigma{}_c$
779: & $t_{last}$ (Gyr)
780: & $m_{\rm BSS}$ ($M_\odot{}$)
781: & $\chi{}_{\rm RGB}^2$
782: & $\chi{}_{\rm HB}^2$\\
783: \hline
784: A1 & 0.8 & 3.5 & 0 & 2 & 1.3 & 0.31 & 0.26 \\
785: A2 & 0.8 & 3.5 & 0 & 1 & 1.3 & 0.54 & 0.47 \\
786: A3 & 0.8 & 3.5 & 0 & 4 & 1.3 & 0.58 & 0.49 \\
787: A4 & 0.8 & 3.5 & 0 & 10 & 1.3 & 0.64 & 0.54 \\
788: B1 & 0.0 & 3.5 & 0 & 2 & 1.3 & 4.51 & 4.08 \\
789: B2 & 0.2 & 3.5 & 0 & 2 & 1.3 & 2.46 & 2.23 \\
790: B3 & 0.5 & 3.5 & 0 & 2 & 1.3 & 0.62 & 0.60 \\
791: B4 & 1.0 & 3.5 & 0 & 2 & 1.3 & 1.09 & 0.94 \\
792: C1 & 0.8 & 4.5 & 0 & 2 & 1.3 & 4.02 & 3.64 \\
793: C2 & 0.8 & 3.0 & 0 & 2 & 1.3 & 1.89 & 1.72 \\
794: C3 & 0.8 & 2.5 & 0 & 2 & 1.3 & 7.03 & 6.47 \\
795: D1 & 0.0 & 4.5 & 0 & 2 & 1.3 & 2.74 & 2.47 \\
796: D2 & 0.0 & 3.0 & 0 & 2 & 1.3 & 8.67 & 7.89 \\
797: E1 & 0.8 & 3.5 & 0 & 2 & 1.1 & 0.31 & 0.26 \\
798: E2 & 0.8 & 3.5 & 0 & 2 & 1.5 & 0.56 & 0.47 \\
799: F1 & 0.8 & 3.5 & 1.& 2 & 1.3 & 0.85 & 0.73 \\
800:
801: \noalign{\vspace{0.1cm}}
802: \hline
803: \end{tabular}
804: \end{center}
805: \footnotesize{
806: $\chi{}_{\rm RGB}^2$ ($\chi{}_{\rm HB}^2$) indicates the $\chi{}^2$ of the $N_{{\rm BSS}}/N_{{\rm RGB}}$ ($N_{{\rm BSS}}/N_{{\rm HB}}$) distribution. The reported values of $\chi{}_{\rm RGB}^2$ and $\chi{}_{\rm HB}^2$ are not reduced and have been calculated on the basis of 5 data points.
807: }
808: \end{table*}
809: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
810:
811:
812: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% TABLE 5%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
813: \begin{table*}
814: \begin{center}
815: \caption{Simulation parameters and $\chi{}^2$ for Ursa Minor.}
816: \leavevmode
817: \begin{tabular}[!h]{llllllll}
818: \hline
819: Run
820: & $r_{min}/r_c$
821: & $r_{max}/r_c$
822: & $v_{kick}/\sigma{}_c$
823: & $t_{last}$ (Gyr)
824: & $m_{\rm BSS}$ ($M_\odot{}$)
825: & $\chi{}_{\rm RGB}^2$
826: & $\chi{}_{\rm HB}^2$\\
827: \hline
828: A1 & 0.5 & 1.9 & 0 & 2 & 1.3 & 0.42 & 0.36 \\
829: A2 & 0.5 & 1.9 & 0 & 1 & 1.3 & 0.32 & 0.28 \\
830: A3 & 0.5 & 1.9 & 0 & 4 & 1.3 & 0.20 & 0.17 \\
831: A4 & 0.5 & 1.9 & 0 & 10 & 1.3 & 0.34 & 0.29 \\
832: B1 & 0.0 & 1.9 & 0 & 2 & 1.3 & 1.68 & 1.50 \\
833: B2 & 0.2 & 1.9 & 0 & 2 & 1.3 & 0.58 & 0.53 \\
834: B3 & 0.8 & 1.9 & 0 & 2 & 1.3 & 1.93 & 1.61 \\
835: B4 & 1.0 & 1.9 & 0 & 2 & 1.3 & 3.51 & 2.95 \\
836: C1 & 0.5 & 3.0 & 0 & 2 & 1.3 & 18.07 & 16.06 \\
837: C2 & 0.5 & 2.5 & 0 & 2 & 1.3 & 4.79 & 4.24 \\
838: C3 & 0.5 & 1.0 & 0 & 2 & 1.3 & 6.99 & 6.21 \\
839: D1 & 0.0 & 3.0 & 0 & 2 & 1.3 & 12.60 & 11.29 \\
840: D2 & 0.0 & 1.5 & 0 & 2 & 1.3 & 5.53 & 4.89 \\
841: E1 & 0.5 & 1.9 & 0 & 2 & 1.1 & 0.42 & 0.36 \\
842: E2 & 0.5 & 1.9 & 0 & 2 & 1.5 & 0.43 & 0.36 \\
843: F1 & 0.5 & 1.9 & 1 & 2 & 1.3 & 2.67 & 2.33 \\
844: \noalign{\vspace{0.1cm}}
845: \hline
846: \end{tabular}
847: \end{center}
848: \footnotesize{
849: $\chi{}_{\rm RGB}^2$ ($\chi{}_{\rm HB}^2$) indicates the $\chi{}^2$ of the $N_{{\rm BSS}}/N_{{\rm RGB}}$ ($N_{{\rm BSS}}/N_{{\rm HB}}$) distribution. The reported values of $\chi{}_{\rm RGB}^2$ and $\chi{}_{\rm HB}^2$ are not reduced and have been calculated on the basis of 6 data points.
850: }
851: \end{table*}
852: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
853:
854: Initial velocities are generated from the distributions described in
855: Sigurdsson \& Phinney (1995). In most runs, no initial kicks
856: are given to BSSs, because they are expected to be
857: MT-BSSs. We also made some (physically unrealistic) check run, were a
858: kick velocity ($v_{kick}$) is given to BSSs born inside the core.
859:
860: In most of the runs the mass of the BSSs is assumed to be $m_{\rm
861: BSS}=1.3\,{}M_\odot{}$. We made check runs with masses in the range from
862: 1.1 to 1.5 $M_\odot{}$ (higher masses are unlikely, at least for some
863: globular clusters; see Ferraro et al. 2004, 2006). This range of masses
864: is also consistent with the isochrones for our data of Draco and Ursa
865: Minor (see Appendix B).
866: %, without noting substantial differences. CONTROLLARE DI NUOVO!
867:
868: Each BSS is evolved for a time $t$, randomly selected from a
869: homogeneous distribution between $t=0$ and $t=t_{last}$. The parameter
870: $t_{last}$ represents the lifetime of BSSs (see M04, M06). We made runs
871: with $t_{last}$=1, 2, 4, 10 Gyr.
872: %The results are quite similar, since dynamical friction is quite
873: %inefficient in dSphs. Thus, we cannot estimate the age of BSS in dSphs
874: %merely on the basis of dynamics. CONTROLLARE DI NUOVO!!!!
875:
876: %dN(r)\propto{}r^{-2}
877:
878:
879: \subsection{Comparison with observations}
880: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% FIGURE 6 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
881: \begin{figure*}
882: \center{{
883: \epsfig{figure=fig6a.eps,height=8cm}
884: \epsfig{figure=fig6b.eps,height=8cm}
885: }}
886: \caption{\label{fig:fig7}
887: Relative frequency of BSSs normalized to RGB (left panel) and HB stars (right panel) in Draco. The filled circles connected by the solid line are the measurements (the same as in Fig.~\ref{fig:fig3}). The open circles connected by the dashed line are the fiducial model (run A1).
888: }
889: \end{figure*}
890: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
891: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% FIGURE 7 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
892: \begin{figure*}
893: \center{{
894: \epsfig{figure=fig7a.eps,height=8cm}
895: \epsfig{figure=fig7b.eps,height=8cm}
896: }}
897: \caption{\label{fig:fig8}
898: Relative frequency of BSSs normalized to RGB (left panel) and HB stars (right panel) in Ursa Minor. The filled circles connected by the solid line are the measurements (the same as in Fig.~\ref{fig:fig3}). The open circles connected by the dashed line are the fiducial model (run A1).
899: }
900: \end{figure*}
901: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
902:
903: We ran different simulations, adopting different masses and lifetimes
904: for BSSs, and varying the interval [$r_{min}$, $r_{max}$] where MT-BSSs
905: are allowed to form. For each of them, we obtain the
906: distributions $N_{\rm BSS}/N_{\rm RGB}$ and $N_{\rm BSS}/N_{\rm HB}$,
907: and calculate their $\chi{}^2$ ($\chi{}_{\rm RGB}^2$ and $\chi{}_{\rm
908: HB}^2$, respectively) with respect to observations.
909:
910: From the $\chi{}^2$ analysis (Tables 4 and 5) it appears that the
911: lifetime of BSSs, $t_{last}$, does not affect the results: runs A1, A2,
912: A3 and A4, which differ only for $t_{last}$,
913: %($t_{last}$=1, 2, 4 and 10 Gyr, respectively),
914: have $\chi{}_{\rm RGB}^2\sim{}\chi{}_{\rm
915: HB}^2\sim{}1$, both in Draco and in Ursa Minor. Then, in the case of
916: dSphs, our simulations cannot constrain the age of BSSs. The reason
917: is that dSphs are dynamically 'quiet' environments, where, due to the low
918: density, both dynamical friction and close interactions are inefficient.
919:
920: On the other hand, if BSSs burn a tiny amount of hydrogen, acquired from
921: the companion stars, they are expected to be relatively short lived.
922: Thus, the run with $t_{last}=10$ Gyr (A4 for both Draco and Ursa Minor)
923: is likely unrealistic. In the following, we will consider $t_{last}=2$
924: Gyr as the fiducial value, for analogy with the findings of M04 and M06
925: for globular clusters, and because an age of about $2$ Gyr is suggested
926: also by isochrones (see Appendix~B)
927: %Ferraro et
928: %al. 1993, 1997 {\bf [MM: MA QUI SAREBBE MEGLIO SE CI FACESSIMO DELLE
929: %ISOCHRONE DI NOSTRO PUGNO] [ER: ??? ne parliamo a voce] [MM: devi guardare dove va a finire la massa di turn off dell'isocrona che fitta meglio le BSS assumendo che siano MS: questa ti da' all'incirca la massa]}).
930:
931:
932: Also the mass of the BSS is not a crucial parameter: runs A1, E1 and E2,
933: which differ only in the BSS mass,
934: %($m_{\rm BSS}=$ 1.3, 1.1 and 1.5 $M_\odot{}$, respectively),
935: have $\chi{}_{\rm RGB}^2\sim{}\chi{}_{\rm
936: HB}^2\lesssim{}1$. In most of the runs, we assume as fiducial value $m_{\rm
937: BSS}=1.3\,{}M_\odot{}$. Masses larger than $m_{\rm
938: BSS}\sim{}1.4-1.5\,{}M_\odot{}$ tend to be discarded by observations,
939: both in our data (see Appendix~B) and in globular clusters (Ferraro et
940: al. 2006).
941:
942: The parameters which mainly affect our results are the lower and upper
943: limit of the initial radial position distribution ($r_{min}$ and
944: $r_{max}$). We remind that initial positions in such simulations
945: represent the point where a binary which is undergoing mass transfer
946: turns into a BSS.
947:
948: The best-fitting value for $r_{min}$ is similar for Draco and Ursa
949: Minor, and is equal to 0.8 $r_c$ and 0.5 $r_c$, respectively. All the
950: values of $r_{min}$ from 0 to $\sim{}1\,{}r_c$ give acceptable
951: $\chi{}^2$ (see e.g. runs B1$-$B4 both for Draco and for Ursa Minor).
952:
953: The best-fitting $r_{max}$ (expressed in terms of $r_c$) is a factor of
954: $\sim{}2$ larger for Draco (3.5 $r_c$) than for Ursa Minor (1.9
955: $r_c$). Indeed, it is possible to reproduce Draco BSSs, recovering an
956: acceptable $\chi{}^2$, also with $2.5\leq{}r_{max}/r_c\leq{}4.5$,
957: whereas in the case of Ursa Minor $r_{max}>2.5\,{}r_c$ and
958: $r_{max}<\,{}r_c$ are inconsistent with observations (see e.g. runs
959: C1$-$C3 both for Draco and for Ursa Minor).
960:
961: This discrepancy might be due only to the different normalization. In
962: fact, the core radius of Ursa Minor is approximately twice as large as
963: that of Draco. In physical units, the best fits are
964: $r_{min}\sim{}370$ and $r_{max}\sim{}1600$ arcsec (150 and 640 pc) for
965: Draco, and $r_{min}\sim{}470$ and $r_{max}\sim{}1800$ arcsec (160 and
966: 620 pc) for Ursa Minor.
967:
968: %Furthermore, the real
969: %extension of the tidal radius of Ursa Minor and especially of Draco is
970: %still under debate. For $r_t$ we adopt the conservative values by IH95;
971: % but there are claims that the tidal radius of
972: %Draco is a factor $\sim{}1.5$ larger (A01;
973: %Piatek et al. 2001).
974:
975: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% FIGURE 4 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
976: %\begin{figure*}
977: %\center{{
978: %\epsfig{figure=lee.eps,height=8cm}
979: %\epsfig{figure=lee_ursa_nofg.eps,height=8cm}
980: %}}
981: %\caption{\label{fig:figlee}
982: %Observed relative frequency of BSS normalized to SGB stars. Left panel refers to Draco, right panel to Ursa Minor. The solid line refers to all BSS, the dotted line to faint BSS and the dashed line to bright BSS. BSS are defined [accordingly to L03] as stars with $20.9<$V$<22.8$ and with $-0.1<$(V-I)$<0.2$ for Draco and $0.1<$(V-I)$<0.4$ for Ursa Minor. SGB are RGB stars with $20.9<$V$<22.8$. Faint (bright) BSS are BSS with $22.1<$V$<22.8$ ($20.9<$V$<22.1$).
983: %}
984: %\end{figure*}
985: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
986:
987:
988: The runs labelled as F1 in the case of both Draco and Ursa Minor have
989: been set up by taking the best-fitting parameters (runs labelled as A1)
990: and adding a small kick velocity $v_{kick}=\sigma{}_c$ to BSSs born inside $r_c$. This check is
991: physically unrealistic, as the natal kick is associated with COL-BSSs,
992: which cannot form in dSphs. Interestingly, the $\chi{}^2$ is quite good
993: in both cases. However, we note that more than 10 per cent of BSSs are
994: 'spuriously' ejected in these runs.
995:
996:
997: Figs.~\ref{fig:fig7} and \ref{fig:fig8} compare our fiducial
998: model (run A1) with observations, for Draco and Ursa Minor,
999: respectively. The good agreement with data is evident: the model has
1000: $\chi{}^2_{RGB}=\chi{}^2_{HB}\sim{}0.3$ for Draco (5 data points) and
1001: $\chi{}^2_{RGB}=\chi{}^2_{HB}\sim{}0.4$ for Ursa Minor (6 data points).
1002: %In Figs.~7 and 8 we report also the comparison with a flat distribution of $N_{\rm BSS}/N_{\rm RGB}$ and $N_{\rm BSS}/N_{\rm HB}$.
1003:
1004: In summary, the dynamical simulations reproduce the
1005: observations very well
1006: for all the possible MT-BSS masses and lifetimes in the
1007: range allowed by the models. The best fit is achieved for the model with
1008: $r_{min}=0.8\,{}r_c$ and $r_{max}=3.5\,{}r_c$ for Draco, and with
1009: $r_{max}=0.5\,{}r_c$ and $r_{max}=1.9\,{}r_c$ for Ursa Minor.
1010: However, all $r_{min}$ from 0 to $r_c$ are acceptable, as
1011: well as all the $r_{max}$ within $\approx{}0.5\,{}r_c$ from the
1012: best-fitting value. Thus, BSS candidates are consistent with a
1013: population initially distributed in an
1014: isothermal sphere between the centre of the galaxy and the tidal
1015: radius. This result agrees with the model of BSS formation from
1016: mass-transferring binaries, and hints that BSS candidates in Draco and Ursa Minor are real MT-BSSs.
1017:
1018:
1019: \section{Comparison with other galaxies and globular clusters}
1020:
1021: BSSs have been observed in most globular clusters and at least in four
1022: dSphs: Draco (A01), Sculptor (Hurley-Keller et
1023: al. 1999), Sextans (L03), and Ursa Minor (C02).
1024: It is instructive to compare our findings with previous papers on both
1025: dSphs and globular clusters.
1026:
1027:
1028: \subsection{Comparison with other dSphs}
1029:
1030: Previous studies of Draco (A01) and Sculptor
1031: (Hurley-Keller et al. 1999) do not report information about the radial
1032: distribution or the luminosity of BSSs. However, Hurley-Keller et
1033: al. (1999) calculate the ratio $N_{\rm BSS}/N_{\rm SGB}$ in the inner
1034: region of the galaxy (a box of 15'$\times{}$15' centred on the centre of the galaxy) and in the outer
1035: one. They find that this ratio changes only by a factor 1.5 between the
1036: inner and the outer region, suggesting that BSSs are not very
1037: concentrated.
1038:
1039: On the other hand, A01 suggest that BSS candidates in Draco are
1040: consistent with a population of intermediate age stars. In fact, they
1041: find a 'red clump' population in the CMD diagram, which might support
1042: this interpretation. This hypothesis cannot be ruled out also in the
1043: case of our data (see discussion in Appendix~B).
1044: % which hypothesis ? this is a bit unclear
1045:
1046: C02 analysed the ratio of the number of `blue plume' stars (corresponding
1047: to a wider definition of BSS) and the number of HB stars as a function of
1048: radius (see their fig.~10) in Ursa Minor. They find an almost flat
1049: distribution, whereas we note a small rise in the relative frequency of BSSs
1050: around $1.4\,{}r_c$, but given the large error bars, our
1051: distribution is also consistent with a flat one ($\chi{}^2\sim{}1$
1052: %marginally ?
1053: with 6 data points). However, not only was the definition of BSS in C02
1054: different, but also the observed photometric bands: C02 build their CMD
1055: by plotting a {\it `V'} magnitude, which is actually the average
1056: between $R$ and $B$ magnitudes, versus the ($B-R$) colour. Thus, our results
1057: and those of C02 are not directly comparable. The most important fact is that
1058: both C02 and our findings suggest that there is no central peak of BSSs
1059: in Ursa Minor.
1060:
1061: As we already mentioned in Section 3, L03 show both the radial and the
1062: luminosity distribution of BSS candidates in Sextans. Furthermore, their
1063: data are more easily comparable with ours, as they use the same
1064: filters. However, even if we adopt the same definition of BSS and the
1065: same normalization as L03, we do not find in either Draco or Ursa Minor
1066: any correlation between the brightness and the radial position of the BSSs,
1067: unlike that reported by L03 in Sextans.
1068: This discrepancy is unlikely due to the lack of statistics in our data, because, when we adopt the same selection criteria as L03, the number of BSSs rises to 198 in Draco and 212 in Ursa Minor, which is quite close to the sample of L03 (i.e. $\sim{}~230$ BSSs).
1069: %(Fig.~\ref{fig:figlee}).
1070:
1071: %In Fig.~\ref{fig:figboh} we report the BSS radial distribution in Draco (left) and Ursa Minor (right panel), adopting the L03
1072: %definition and normalization (but the results do not change if we adopt
1073: %our definitions).
1074: %We do not observe any significant
1075: %differences between faint ($22.1<V<22.8$) and bright ($20.9<V<22.1$)
1076: %BSS.
1077: %Indeed, bright BSS in Draco seem slightly less concentrated than
1078: %the faint ones. The discrepancy between our results and that of L03
1079: %could be explained with an intrinsic difference of BSS in
1080: %Sextans with respect to Draco and Ursa Minor (see Section 5).
1081:
1082: %The distribution of $N_{{\rm BSS}}/N_{{\rm SGB}}$
1083: %%($N_{{\rm SGB}}$ is the number of SGB stars per each annulus)
1084: %is very
1085: %similar to the distribution of both $N_{{\rm BSS}}/N_{{\rm RGB}}$ and
1086: %$N_{{\rm BSS}}/N_{{\rm HB}}$. These properties suggest
1087:
1088:
1089:
1090:
1091: The difference here might be due to statistical fluctuations, or could
1092: simply be connected with the intrinsic properties of Sextans, which are quite
1093: different from those of Draco and Ursa Minor. For example, Sextans has a
1094: higher concentration index ($c\sim{}1$) with respect to both Draco and Ursa
1095: Minor, and it is very extended ($r_c$= 16'.6 and $r_t=160$', IH95). L03
1096: show the radial distribution of BSSs only within $\sim{}1.1\,{}r_c$,
1097: without information about external BSSs. It has recently
1098: been claimed that the centre of Sextans contains a
1099: kinematically distinct stellar population, which might be associated
1100: with a star cluster (Kleyna et al. 2004; but also see Walker et al. 2006).
1101: Indeed, the correlation between position and brightness of BSSs could be
1102: explained by invoking the presence of a star cluster. In this case, the bright
1103: BSSs, more concentrated toward the centre, could be COL-BSSs or even young
1104: stars formed in
1105: the star cluster; whereas the faint BSSs are MT-BSSs, like those in Draco
1106: and Ursa Minor. (We note that the distribution of faint BSSs alone in
1107: Sextans is quite similar to the distribution of the entire BSS sample in
1108: Draco and Ursa Minor.)
1109:
1110:
1111: \subsection{Comparison with globular clusters}
1112: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% FIGURE 8 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1113: \begin{figure}
1114: \center{{
1115: \epsfig{figure=fig8.eps,height=8.5cm}
1116: }}
1117: \caption{\label{fig:fig9}
1118: From left to right and top to bottom: relative frequency of BSSs normalized to HB stars in 47~Tucanae, $\omega{}$ Centauri, Draco and Ursa Minor. Filled (open) circles connected by the solid (dashed) line are the observations (best-fitting simulations). The dotted vertical line in 47~Tucanae and $\omega{}$ Centauri panels indicate $r_{av}$. The data and models for Draco and Ursa Minor are the same as in Fig.~\ref{fig:fig7} and Fig.~\ref{fig:fig8}, respectively. The data for 47~Tucanae and $\omega{}$ Centauri are from Ferraro et al. (2004) and from Ferraro et al. (2006), respectively. Models for 47~Tucanae and $\omega{}$ Centauri are from M06.
1119: }
1120: \end{figure}
1121: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1122:
1123:
1124: What are the main differences between BSSs in dSphs and BSSs in
1125: globular clusters? Are there any globular clusters whose BSSs behave like
1126: those in dSphs? Fig.~\ref{fig:fig9} shows the distribution of $N_{\rm
1127: BSS}/N_{\rm HB}$ of Draco and Ursa Minor together with that of
1128: 47~Tucanae and $\omega{}$ Centauri (from M06). We chose 47~Tucanae and
1129: $\omega{}$ Centauri, because they have very different BSS populations.
1130:
1131: 47~Tucanae is a sort of prototype for BSSs in globular clusters: it
1132: clearly shows the bimodal BSS relative frequency, which has been
1133: observed in more and more globular clusters in the last few years
1134: (Ferraro et al. 1993, 1997; Zaggia et al. 1997; Ferraro et al. 2004;
1135: Sabbi et al. 2004; Warren, Sandquist \& Bolte 2006; Lanzoni et
1136: al. 2007a). The clusters which do not have a bimodal distribution,
1137: in general show only the central peak of BSSs, which rapidly drops outside
1138: the core (see e.g. NGC1904; Lanzoni et al. 2007b).
1139:
1140: The bimodal distribution might be explained by requiring that the
1141: central peak is populated mainly by COL-BSSs, the external
1142: increase is due to MT-BSSs which have not yet sank to the centre, and
1143: the minimum of the BSS distribution is connected with the efficiency of
1144: dynamical friction (M04, M06). In fact, the position of the minimum has
1145: been found to be equal to the maximum distance ($r_{av}$) from the
1146: centre at which dynamical friction is able to bring binaries
1147: (progenitors of MT-BSSs) into the core within the lifetime of the cluster
1148: (M04, M06). In this scenario, globular clusters without the external
1149: rise (e.g. NGC1904) are expected to be poor in mass-transferring binaries,
1150: or not to have formed MT-BSSs in the last Gyrs.
1151:
1152: From Fig.~\ref{fig:fig9} it is clear that the distribution of BSSs in
1153: Draco and Ursa Minor is completely different from that of a typical
1154: globular cluster like 47~Tucanae. In particular, it seems that Draco and
1155: Ursa Minor have only the peripheral rise of BSSs, and completely lack the
1156: central peak. As we already discussed in Section~3, this supports the
1157: idea that the central peak in globular clusters is due to COL-BSSs,
1158: whereas the external rise is due to MT-BSSs.
1159:
1160: Instead, $\omega{}$ Centauri is unique among the globular
1161: clusters where BSSs have been already observed. In fact, $N_{\rm
1162: BSS}/N_{\rm HB}$ and $N_{\rm BSS}/N_{\rm RGB}$ in $\omega{}$ Centauri
1163: are both consistent with a flat distribution. M06 suggested that this
1164: distribution might be the product of both the lack of COL-BSSs in the
1165: core of $\omega{}$ Centauri (its core density being quite low)
1166: and the inefficiency of dynamical friction. In fact, due to the joint
1167: effect of a low central density ($\sim{}6\times{}10^3$ stars pc$^{-3}$)
1168: and of a high velocity dispersion ($\sim{}17$ km s$^{-1}$), the
1169: dynamical friction time-scale in $\omega{}$ Centauri is a factor of
1170: $\sim{}$200 longer than in 47~Tucanae. As dynamical
1171: friction is inefficient, binaries do not sink into the centre, and the
1172: minimum in the BSS distribution does not appear.
1173:
1174: In this sense, $\omega{}$ Centauri appears as something midway between the other globular clusters and the
1175: dSphs. Different from dSphs, it can still form COL-BSSs in the
1176: centre, because its core density is considerably higher than that of
1177: dSphs; but its dynamical friction is inefficient in
1178: moulding the shape of BSS distribution, exactly as in
1179: dSphs. Fig.~\ref{fig:fig9} even suggests the idea of continuity between
1180: 47~Tucanae, $\omega{}$ Centauri and the two dSphs: as the central
1181: density of the system decreases, the central peak disappears, and the
1182: BSS distribution becomes less and less concentrated.
1183:
1184: In line with this idea is the distribution of $r_{av}$ (see Fig.~\ref{fig:fig9}), which
1185: in 47~Tucanae and in many other globular clusters is $\sim{}10\,{}r_c$,
1186: in $\omega{}$ Centauri is $\sim{}1\,{}r_c$, while in Draco and Ursa
1187: Minor it does not even appear in the plot, because it is consistent with
1188: 0 ($r_{av}\lesssim{}5\times{}10^{-2}\,{}r_c$).
1189:
1190: We note that, apart from the BSS distribution,
1191: $\omega{}$ Centauri displays several features which are indicative of an object
1192: midway between globular clusters and dSphs: the metallicity spread, the evidences for rotation, the large mass and the low concentration are quite atypical for a globular cluster; so that some authors (Zinnecker et al. 1988;
1193: Freeman 1993; Ideta \& Makino 2004) claim that $\omega{}$ Centauri is
1194: not a real globular cluster, but the nuclear remnant of a dwarf galaxy.
1195:
1196: %It is not a case if some authors (Zinnecker et al. 1988;
1197: %Freeman 1993; Ideta \& Makino 2004) claim that $\omega{}$ Centauri is
1198: %not a real globular cluster, but the remnant of a dwarf galaxy.
1199:
1200: Finally, in Fig.~\ref{fig:fig9} it is also apparent that the level of $N_{\rm BSS}/N_{\rm HB}$ ($\sim{}0.05-0.4$) in Draco and Ursa Minor is comparable with the level in 47~Tucanae and $\omega{}$ Centauri, analogous to most of the globular clusters (see e.g. M06). Thus, we can conclude that the fraction of BSSs versus HB stars in these two dSphs and in globular clusters are similar. This fact indirectly supports the hypothesis that BSS candidates in Ursa Minor and Draco are real MT-BSSs. In fact, if all the BSSs are MT-BSSs, $N_{\rm BSS}/N_{\rm HB}$ should
1201: reflect both the ratio of lifetimes and the fraction of stars in suitable
1202: binaries,
1203: and should be constant for the same turn-off mass and metallicity populations, if
1204: the binary fraction is constant.
1205: %Interestingly, $\omega{}$ Centauri is different from the other clusters, with an uncommonly low $N_{\rm BSS}/N_{\rm HB}$ (Ferraro et al. 2006).
1206:
1207: \section{Summary}
1208:
1209: In this paper we addressed the problem of BSS candidates in
1210: dSphs in general, and in Draco and Ursa Minor in particular. There are
1211: two fundamental open questions about BSSs in dSphs: i) whether
1212: they are authentic BSSs or young stars; and ii) what is their formation
1213: mechanism?
1214:
1215: We analysed both the radial and the luminosity distributions of these
1216: stars, and we compared the data with dynamical simulations of BSSs. The
1217: main feature of the observed radial distribution of BSSs, normalized to
1218: RGB or HB stars, is the absence of a central peak. Even if the young
1219: % ??
1220: stars' interpretation cannot be dismissed (at least for Draco; see
1221: Appendix~B), this suggests that BSS candidates in Draco and Ursa Minor
1222: are actually true BSSs. Furthermore, the almost flat radial distribution
1223: is consistent with theoretical models (M06) for MT-BSSs, i.e. BSSs which
1224: formed by mass transfer in isolated binaries. Also the luminosity
1225: distribution, which does not show any correlation with the position of
1226: BSSs, agrees with theoretical models of mass-transfer BSS formation.
1227: %Finally, we found that the observed radial distribution agrees very well
1228: %with dynamical simulations of MT-BSS.
1229:
1230: These findings support the model by M04 and M06, which
1231: explains the formation of BSSs by the joint contribution of stellar
1232: collisions and mass transfer in isolated binaries. This model was
1233: originally developed only for globular
1234: clusters, but we find that it works also for dSphs. As predicted by M06,
1235: the presence of a central peak in the relative frequency of BSSs is due
1236: to COL-BSSs, and can be explained only if both stellar collisions and
1237: dynamical friction are efficient. The peak tends to disappear if the
1238: central density of the system is too low and/or its dynamical friction
1239: time is too long. This idea was confirmed by the absence of any
1240: central peak in $\omega{}$ Centauri, and now we find that this
1241: result is even stronger in Draco and Ursa Minor.
1242:
1243: Low-density systems, where stellar collisions do not occur, can form
1244: only MT-BSSs, whose initial distribution mirrors the distribution of the
1245: progenitor binaries. The less efficient the dynamical friction, the more
1246: the BSS distribution is similar to the distribution of progenitor
1247: binaries. This idea is
1248: fully supported by Draco and Ursa Minor BSSs: the best-fitting
1249: simulations are based on an isothermal distribution between
1250: (approximately) the core and the tidal radius, as we would expect for a
1251: distribution of primordial binaries.
1252:
1253: Furthermore, Momany et al. (2007) recently analysed the BSS candidates of 8 dSphs (Draco and Ursa Minor among them) and found a statistically significant anti-correlation between the relative frequency of BSS candidates ($N_{\rm BSS}/N_{\rm HB}$, calculated over the entire galaxy) and the total luminosity of the dSph. If BSS candidates were young MS stars rather than real BSSs, such anti-correlation would not make sense.
1254:
1255: Thus, from our analysis as well as from Momany et al. (2007) we conclude that BSS candidates in Draco and Ursa Minor behave like real MT-BSSs, rather than young MS stars. This suggests (even if it does not definitely prove) that Draco and Ursa Minor are 'fossil' galaxies, where star formation was completely
1256: suppressed many Gyrs ago. This scenario is also confirmed by recent
1257: simulations (Mayer et al. 2007), which indicate that Draco and Ursa
1258: Minor, two of the closest dSphs to the Milky Way,
1259: had all their gas removed $\sim{}10$~Gyr ago, probably by tidal shocks and
1260: ram pressure
1261: exerted
1262: by the Milky Way. The 'fossil' nature of Draco and Ursa Minor would make
1263: them a natural place to study the conditions at the earliest epochs of galaxy formation.
1264:
1265:
1266: On the other hand, it would be interesting to throughly study the
1267: nature of BSS candidates in other dSphs, like Sextans,
1268: where star formation probably lasted longer. The main goal would be
1269: to understand whether, and what fraction of, these stars are authentic
1270: BSSs, in order to disentangle the history of BSS formation from that of
1271: MS stars.
1272:
1273: \section {Acknowledgments}
1274: We thank the referee, T. Maccarone, for the critical reading of the
1275: manuscript. We also thank F. D'Antona, F. Ferraro, S. Zaggia and
1276: Y. Momany for useful discussions. MM acknowledges support from the Swiss
1277: National Science Foundation, project number 200020-109581/1
1278: (Computational Cosmology \&{} Astrophysics). ER acknowledges support from the
1279: Netherlands Organization for Scientific Research (NWO) under project
1280: number 436016. MM and ER thank the Kapteyn Astronomical Institute of the
1281: University of Groningen, and
1282: the Institute for Theoretical Physics of the University of Z\"urich for
1283: the hospitality during the preparation of this paper.
1284:
1285:
1286: \begin{thebibliography}{}
1287:
1288: \bibitem{}Aparicio A., Carrera R., Mart\'inez-Delgado D., 2001, AJ, 122,
1289: 2524 [A01]
1290:
1291: %\bibitem{a} Bailyn C. D., 1995, ARA\&{}A, 33, 133
1292:
1293: \bibitem{}Bailyn C. D., Pinsonneault M. H., 1995, ApJ, 439, 705
1294:
1295: \bibitem{}Battaglia G. et al., 2006, A\&{}A, 459, 42
1296:
1297: \bibitem{}Bellazzini M., Ferraro F.R., Origlia L., Pancino E., Monaco
1298: L., Oliva E., 2002, AJ 124, 3222
1299:
1300:
1301: %\bibitem{} Beccari, G., Ferraro, F.R., Possenti, A., Valenti, E., Origlia, L., Rood, R.T., 1996, AJ, 131, 2551
1302:
1303: \bibitem{}Beers T. C., Almeida T., Rossi S., Wilhelm R., Marsteller B., 2007, ApJS, 168, 277
1304:
1305: \bibitem{}Bellazzini M., Ferraro F. R., Origlia L., Pancino E., Monaco L., Oliva E., 2002, AJ, 124, 3222
1306:
1307: %\bibitem{c} Bellazzini M., Ferraro F. R., Sollima A., Pancino E., Origlia L., 2004, A\&{}A, 424, 199
1308:
1309: %\bibitem{d} Bellazzini M., Fusi Pecci F., Messineo M., Monaco L., Rood R. T.,2002, AJ, 123, 1509
1310:
1311: \bibitem{}Benz W., Hills J. G., 1987, ApJ, 323, 614
1312:
1313: \bibitem{}Benz W., Hills J. G., 1992, ApJ, 389, 546
1314:
1315:
1316: %\bibitem{e} Binney J., Tremaine S., 1987, Galactic Dynamics (Princeton: Princeton Univ. Press)
1317:
1318: \bibitem{}Carney B. W., Latham D. W., Laird J. B., Grant C. E., Morse J. A., 2001, AJ, 122, 3419
1319:
1320: \bibitem{}Bonanos A.Z., Stanek K.Z., Szentgyorgyi A.H., Sasselov D.D.,
1321: Bakos G.Á., 2004, AJ, 127, 861
1322:
1323:
1324:
1325: \bibitem{}Carney B. W., Latham D. W., Laird J. B., 2005, AJ, 129, 466
1326:
1327: \bibitem{}Carrera R., Aparicio A., Mart\'inez-Delgado D.,
1328: Alonso-Garc\'ia J., 2002, AJ, 123, 3199 [C02]
1329:
1330: \bibitem{}Chabrier G., 2003, PASP, 115, 763
1331:
1332: %\bibitem{g} Colpi M., Mapelli M., Possenti A., 2003, ApJ, 599, 1260
1333: %\bibitem{h} D'Amico N., et al., 2002, ApJ, 570, L89
1334:
1335: \bibitem{}Davies M. B., Benz W., Hills J. G., 1994, ApJ, 424, 870
1336:
1337: \bibitem{} Davies M. B., Piotto G., De Angeli F., 2004, MNRAS, 349, 129
1338:
1339: %\bibitem{j} Drukier G. A., Bailyn C. D., Van Altena W. F., Girard T. M., 2003, AJ, 125, 2559
1340:
1341: %\bibitem{k} Dubath P., Meylan G., Mayor M., 1997, A\&{}A, 324, 505
1342:
1343: \bibitem{}Ferraro F. R., Fusi Pecci F., Cacciari C., Corsi C., Buonanno R., Fahlman G. G., Richer H. B., 1993, AJ, 106, 2324
1344:
1345: \bibitem{} Ferraro F. R., Paltrinieri, B., Fusi Pecci, F.,Cacciari, C., Dorman, B., rood, R.T., 1997, ApJ, 484, L145
1346:
1347:
1348: \bibitem{}Ferraro F. R., Sills A., Rood R. T., Paltrinieri B., Buonanno R., 2003, ApJ, 588, 464
1349:
1350: \bibitem{}Ferraro F. R., Beccari G., Rood R. T., Bellazzini M., Sills A., Sabbi E., 2004, ApJ, 603, 127
1351:
1352: \bibitem{} Ferraro F. R., Sollima A., Rood R. T., Origlia L., Pancino E., Bellazzini M., 2006, ApJ, 638, 433
1353:
1354: %\bibitem{m} Ferraro F. R., Carretta E., Corsi C. E., Fusi Pecci F., Cacciari C., Buonanno R., Paltrinieri B., Hamilton D., 1997, A\&{}A, 320, 757
1355:
1356:
1357:
1358: %\bibitem{} Ferraro F. R., Messineo, M., Fusi Pecci F., De Palo, M.A., Straniero, O., Chieffi, A., Limongi, M., 1999, AJ, 118, 1738
1359:
1360:
1361:
1362:
1363: %\bibitem{o} Ferraro F. R., Possenti A., Sabbi E., Lagani~P., Rood~R.~T., D'Amico~N., Origlia~L., 2003b, ApJ, 595, 179
1364:
1365:
1366:
1367:
1368: %\bibitem{q} Ferraro F. R., Sollima A., Rood R. T., Origlia L., Pancino E., Bellazzini M., 2006a, ApJ, 638, 433
1369:
1370: %\bibitem{q} Ferraro F. R. et al., 2006b, ApJ, 647, L53
1371:
1372:
1373: \bibitem{}Freeman K. C., 1993, The globular clusters-galaxy connection. Astronomical Society of the Pacific Conference Series, Volume 48, Proceedings of the 11th Santa Cruz Summer Workshop in Astronomy and Astrophysics, held July 19-29, 1992, at the University of California, Santa Cruz, San Francisco: Astronomical Society of the Pacific (ASP), edited by Smith G. H., Brodie J. P., 608
1374:
1375: \bibitem{}Freitag M., Benz W., 2005, MNRAS, 358, 1133
1376:
1377: \bibitem{}Fusi Pecci F., Ferraro F. R., Corsi C. E., Cacciari C., Buonanno R., 1992, AJ, 104, 1831
1378:
1379: %\bibitem{s} Gendre B., Barret D., Webb N. A., 2003, A\&{}A, 400, 521
1380:
1381: %\bibitem{t} Gilliland R. L., Bono G., Edmonds P. D., Caputo F., Cassisi S., Petro L. D., Saha A., Shara M. M., 1998, ApJ, 507, 818
1382:
1383: \bibitem{}Girardi L., Bertelli G., Bressan A., Chiosi C., Groenewegen
1384: M.A.T., Marigo P., Salasnich B., Weiss A., 2002, A\&A, 391, 195
1385:
1386: \bibitem{}Harbeck D., Grebel E. K., Holtzman J., Geisler D., Sarajedini A., 2002, in Modes of Star Formation and the Origin of Field Populations, ASP Conference Proceedings, 285, 230. Edited by Grebel E. K. and Brandner W..
1387:
1388: %\bibitem{}Harris W.E., 1996, AJ, 112, 1487
1389:
1390: \bibitem{}Hernandez X., Gilmore G., Valls-Gabaud D., 2000, MNRAS, 317, 831
1391:
1392: \bibitem{}Hurley J. R., Tout C. A., Aarseth S. J., Pols O. R., 2001, MNRAS, 323, 630
1393:
1394: \bibitem{}Hurley J. R., Pols O. R., Aarseth S. J., Tout C. A., 2005, MNRAS, 363, 293
1395:
1396:
1397: \bibitem{}Hurley-Keller D., Mateo M., Grebel E. K., 1999, ApJ, 523L, 25
1398:
1399:
1400: \bibitem{}Ideta M., Makino J.,2004, ApJ, 616L, 107
1401:
1402: \bibitem{}Irwin M.J., 1985, MNRAS, 214, 575
1403:
1404: \bibitem{}Irwin M. J., 1996, in Rodr\'iguez Espinosa J. M., Herrero A., S\'anchez F. (Eds.), Instrumentation for Large Telescopes, VII Canary Islands Winter School, Cambridge University Press, 35
1405: %1996, in Espinosa J., ed., 7th Canary Islands Winter School
1406:
1407: \bibitem{}Irwin M., Hatzidimitriou D., 1995, MNRAS, 277, 1354 [IH95]
1408:
1409: \bibitem{}Irwin M.J., Lewis J., 2001, New Astronomy, 45, 105
1410:
1411: %\bibitem{w} Ivanova N., Belczynski K., Fregeau J. M., Rasio F. A.,
1412: %2005, MNRAS, 358, 572
1413:
1414:
1415: \bibitem{}Kleyna J. T., Wilkinson M. I., Evans N. W., Gilmore G., 2004, MNRAS, 354L, 66
1416:
1417: \bibitem{}Kleyna J.T., Wilkinson M.I., Gilmore G., Evans N.W., 2003,
1418: ApJ, 588, L21
1419:
1420: \bibitem{}Landolt A.U., 1992, AJ, 104, 340
1421:
1422: \bibitem{}Lanzoni B., Dalessandro E., Ferraro F. R., Mancini C., Beccari G., Rood R. T., Mapelli M., Sigurdsson S., 2007a, ApJ, 663, 267
1423:
1424: \bibitem{}Lanzoni B., Sanna N., Ferraro F. R., Valenti E., Beccari G., Schiavon R. P., Rood R. T., Mapelli M., Sigurdsson S, 2007b, ApJ, 663, 1040
1425:
1426: \bibitem{}Lee M. G. et al., 2003, AJ, 126, 2840 [L03]
1427:
1428: \bibitem{}Leonard P. J. T., 1989, AJ, 98, 217
1429:
1430: \bibitem{} Lombardi J. C. Jr., Warren J. S., Rasio F. A., Sills A., Warren A. R., 2002, ApJ, 568, 939
1431:
1432: \bibitem{}Mapelli M., Sigurdsson S., Colpi M., Ferraro F. R., Possenti
1433: A., Rood R. T., Sills A., Beccari G., 2004, ApJ, 605L, 29 [M04]
1434:
1435: \bibitem{}Mapelli M., Sigurdsson S., Ferraro F. R., Colpi M., Possenti
1436: A., Lanzoni B., 2006, MNRAS, 373, 361 [M06]
1437:
1438: \bibitem{}Mateo M., Nemec J., Irwin M., McMahon R., 1991, AJ, 101, 892
1439:
1440: \bibitem{}Mateo M., Fischer P., Krzeminski W., 1995, AJ, 110, 2166
1441:
1442: \bibitem{}Mateo M. L., 1998, ARA\&{}A, 36, 435
1443:
1444: \bibitem{}Mayer L., Kazantzidis S., Mastropietro C., Wadsley J., 2007, Nature, 445, 738
1445:
1446: \bibitem{}McCrea W. H., 1964, MNRAS, 128, 147
1447:
1448: \bibitem{}McMahon R.G., Walton N.A., Irwin M.J., Lewis J.R., Bunclark
1449: P.S., Jones D.H., 2001, New Astronomy, 45, 97
1450:
1451: %\bibitem{z} Merritt D., Meylan G., Mayor M., 1997, AJ, 114, 1074
1452:
1453: %\bibitem{A} Meylan G., Mayor M., Duquennoy A., Dubath P., 1995, A\&{}A, 303, 761
1454: \bibitem{}Momany Y., Held E. V., Saviane I., Zaggia S., Rizzi L., Gullieuszik M., 2007, A\&{}A, 468, 973
1455:
1456: \bibitem{}Monkiewicz J. et al., 1999, PASP, 111, 1392
1457:
1458: \bibitem{}Monkman E., Sills A., Howell J., Guhathakurta P., de Angeli F., Beccari G., 2006, ApJ, 650, 195
1459:
1460: \bibitem{}Naylor T., 1998, MNRAS, 296, 339
1461:
1462: %\bibitem{B} Norris J. E., Freeman K. C., Mighell K. J, 1996, ApJ, 462, 241
1463:
1464: \bibitem{}Piatek S., Pryor C., Armandroff T. E., Olszewski E. W., 2001, AJ, 121, 841
1465:
1466: %\bibitem{} Piotto G. et al., 2004, ApJ, 604L, 109
1467:
1468: \bibitem{}Pols O. R., Marinus M., 1994, A\&{}A, 288, 475
1469:
1470: %\bibitem{C} Pooley D., et al., 2003, ApJ, 591, L131
1471:
1472: %%\bibitem{} Portegies Zwart S. F., Makino J., McMillan S. L. W., Hut P, 1999, A\&{}A, 348, 117
1473: %%\bibitem{} Portegies Zwart S. F., Hut P., Verbunt F., 1997, A\&{}A 328, 130
1474:
1475: %\bibitem{D} Procter Sills A., Bailyn C. D., Demarque P., 1995, ApJ, 455, L163
1476:
1477:
1478: %\bibitem{E} Pryor C., Meylan G., 1993, Structure and Dynamics of Globular Clusters. Proceedings of a Workshop held in Berkeley, California, July 15-17, 1992, to Honor the 65th Birthday of Ivan King. Editors, Djorgovski S.G. and Meylan G.; Publisher, Astronomical Society of the Pacific, Vol. 50, 357
1479:
1480: %\bibitem{F} Reijns R. A., Seitzer P., Arnold R., Freeman K. C., Ingerson T., van den Bosch R. C. E., van de Ven G., de Zeeuw P. T., 2006, A\&{}A, 445, 503
1481:
1482: %%\bibitem{G} Rey S. C., Lee Y. W., Ree C. H., Joo J. M., Sohn Y. J., Walker A. R., 2004, AJ, 127, 958
1483: %\bibitem{H} Rubenstein E. P., Bailyn C. D., 1997, ApJ, 474, 701
1484:
1485: \bibitem{}Sabbi E., Ferraro F. R., Sills A., Rood R. T., 2004, ApJ, 617,
1486: 1296
1487:
1488: \bibitem{}Sandage A. R., 1953, AJ, 58, 61
1489:
1490: \bibitem{}S\'egall M., Ibata R.A., Irwin M.J., Martin N.F., Chapman S.,
1491: 2007, MNRAS, 375, 831
1492:
1493: \bibitem{}Sigurdsson S., Davies M. B., Bolte M., 1994, ApJ, 431, L115
1494:
1495: %\bibitem{K} Sigurdsson S., Phinney E. S., 1993, ApJ, 415, 631
1496:
1497: \bibitem{}Sigurdsson S., Phinney E. S., 1995, ApJS, 99, 609
1498:
1499: \bibitem{}Sills A., Bailyn C. D., 1999, ApJ, 513, 428
1500:
1501: \bibitem{}Sills A., Bailyn C. D., Edmonds P. D., Gilliland R. L., 2000, ApJ, 535, 298
1502:
1503: \bibitem{}Sills A., Faber J. A., Lombardi J. C. Jr., Rasio F. A., Warren A. R., 2001, ApJ, 548, 323
1504:
1505: %%\bibitem{}Spitzer L., 1987, Dynamical Evolution of Globular Clusters. Princeton Univ. Press
1506: %\bibitem{O} Suntzeff N. B., Kraft R. P., 1996, AJ, 111, 1913
1507:
1508: %\bibitem{P} van de Ven G., van den Bosch R. C. E., Verolme E. K., de Zeeuw P. T., 2006, A\&{}A, 445, 513
1509:
1510: %\bibitem{Q} van Leeuwen F., Le Poole R. S., Reijns R. A., Freeman K. C., de Zeeuw P. T., 2000, A\&{}A, 360, 472
1511:
1512: \bibitem{}Tian B., Deng L., Han Z., Zhang X. B., 2006, A\&{}A, 455, 247
1513:
1514: \bibitem{}Tolstoy E. et al., 2004, ApJ, 2004, 617L, 119
1515:
1516: \bibitem{}Walker M. G., Mateo M., Olszewski E. W., Pal J. K., Sen B., Woodroofe M., 2006, ApJ, 642L, 41
1517:
1518: \bibitem{}Warren S. R., Sandquist E. L., Bolte M., 2006, ApJ, 648, 1026
1519:
1520: \bibitem{}Wilkinson M. I., Kleyna J. T., Evans N. W., Gilmore G. F., Irwin M. J., Grebel E. K., 2004, ApJ, 611L, 21
1521:
1522: \bibitem{}Zaggia S. R., Piotto G., Capaccioli M., 1997, A\&{}A, 327, 1004
1523:
1524: \bibitem{}Zinnecker H., Keable C. J., Dunlop J. S., Cannon R. D., Griffiths W. K., 1988, The Harlow Shapley Symposium on Globular Cluster Systems in Galaxies. Proceedings of the 126th Symposium of the International Astronomical Union, held in Cambridge, Massachusetts, U.S.A. August 25-29, 1986. Edited by Jonathan E. Grindlay and A. G. Davis Philip. Kluwer Academic Publishers, Dordrecht, 603
1525:
1526: \end{thebibliography}
1527:
1528: \appendix
1529:
1530: \section{Foreground subtraction}
1531:
1532: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% FIGURE A1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1533: \begin{figure}
1534: \center{{
1535: \epsfig{figure=figA1.eps,height=8.5cm}
1536: }}
1537: \caption{\label{fig:cmd_draco_full}
1538: CMD diagram of Draco. As in Fig. \ref{fig:fig1}, the
1539: selection areas for BSS (1), RGB (2), and HB (3) are shown. We also
1540: show the selection boxes for the HB foreground (4) and the VRS (5).}
1541: \end{figure}
1542: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1543:
1544: The contamination of the data due to foreground Milky Way stars (and
1545: also to background objects) is quite evident, especially for the RGB
1546: region (see Figs. \ref{fig:fig1} and \ref{fig:cmd_draco_full}).
1547: %\footnote{{\bf [ER] The CMD diagrams shown in
1548: %Fig. \ref{fig:fig1} appear to have a different level of contamination
1549: %only because of the spatial extent of the available data: for Draco, the
1550: %observations covered an area of the sky which extends well beyond the
1551: %tidal radius; for Ursa Minor, observations are available only within
1552: %about one tidal radius}}
1553: Removing such contamination is
1554: important, especially in the outer regions of the dSphs.
1555:
1556: We note that, although we refer only to `foreground'
1557: removal, the methods outlined below work equally well for the
1558: subtraction of the contamination by compact background objects.
1559:
1560: %, even if
1561: %this operation is not mentioned in previous papers (e.g. A01).
1562:
1563:
1564:
1565: %In order to avoid foreground contamination, first of all we consider foreground all those stars which are outside $r_{ext}=45$' () from the centre of Draco (Ursa Minor). These stars are well outside the tidal radius, $r_t$, whose value can be derived from Table~1. Then, we subtract the foreground component also for stars within $r_{ext}$ by adopting the following method.
1566:
1567: \subsection{RGB and BSS foreground contamination}
1568:
1569: In order to estimate the foreground contamination for RGB stars and BSSs, we
1570: adopt the following method, both in Draco and in Ursa Minor.
1571:
1572: First of all, we assume that all stars redder than ($V-I$)=1.5 and with $V$
1573: in the [19.5, 23] range (hereafter VRS, i.e. very red stars) are in the
1574: foreground. In fact, the VRS region of the CMD (box 5 in
1575: Fig. \ref{fig:cmd_draco_full}) should not be populated by stars
1576: belonging to Draco, nor to Ursa Minor\footnote{We also checked a
1577: more restrictive definition of VRS, i.e. ($V-I$)$\geq$2.0 and the same $V$
1578: magnitude range. No significant difference was found in our results.}.
1579:
1580: Second, we expect that all the stars which are outside $r_t$ do not
1581: belong to the dSph, independent of their colour and magnitude. This is
1582: useful because the Draco data extend well beyond $r_t$, and we can
1583: select a subset of `external' stars, which we define to include all
1584: the stars whose elliptical radial distance from the centre exceeds
1585: $r_{t}$=45.1'.
1586: %\footnote{Ideally, $r_{ext}\simeq r_t$=28.3'. But in the
1587: %elliptical corona between 28.3' and 45' there is still some
1588: %contamination from Draco stars; this was already reported in previous
1589: %studies, even if it is not clear whether we are in presence of
1590: %extra-tidal stars (IH95), or the tidal radius was underestimated
1591: %(e.g. A01 estimate a tidal radius of 42' for Draco).}.
1592:
1593:
1594: We can therefore count the number of VRS, RGB and BSS equivalents\footnote{Here we use the
1595: name of BSSs and RGB just for convenience. These are foreground
1596: stars which happen to have the same colour and magnitude of BSSs and RGB,
1597: respectively.} with $r_{ell}>r_{t}$ ($N_{{\rm VRS},\,{}ext}$, $N_{{\rm
1598: RGB},\,{}ext}$, and $N_{{\rm BSS},\,{}ext}$, respectively), and we
1599: derive the ratios
1600: $f_{{\rm RGB}/{\rm VRS}} = N_{{\rm RGB},\,{}ext}/N_{{\rm VRS},\,{}ext}
1601: \simeq 0.0318\pm0.0025$,
1602: and
1603: $f_{{\rm BSS}/{\rm VRS}} = N_{{\rm BSS},\,{}ext}/N_{{\rm VRS},\,{}ext}
1604: \simeq 0.0031\pm0.0008$.
1605:
1606: These ratios should be independent of position, as we have tested this in two ways.
1607: First of all, we looked for fluctuations of the surface density of VRS
1608: stars as a function of radius. Although small fluctuations are present,
1609: the overall density can be considered constant in the whole Draco field (it is
1610: consistent with a constant value of 0.94 VRS/arcmin$^2$, with reduced
1611: $\chi{}^2\simeq$0.8 over 23 radial bins).
1612: % You could argue that 2 is significantly different from 1 (~4 sigma).
1613: % Extinction varies noticeably over the Draco region have you corrected for
1614: % this doing the tests ? I have an extinction-corrected version of the
1615: % of the catalogue if required.
1616: As a further test, we split the `external' region into its eastern
1617: and western half and checked that there is no statistically significant
1618: difference between the values of $f_{{\rm RGB}/{\rm VRS}}$ and of
1619: $f_{{\rm BSS}/{\rm VRS}}$ which were obtained in the two halves.
1620:
1621: The foreground contamination of the $i$-th elliptical annulus can be
1622: estimated by counting the number $N_{{\rm VRS},i}$ of VRS stars in the
1623: annulus, and converting it into the expected number of foreground
1624: BSSs (RGB stars) through the factor
1625: $f_{{\rm BSS}/{\rm VRS}}$ ($f_{{\rm RGB}/{\rm VRS}}$). Then, the corrected
1626: number of BSSs (RGB stars) is simply
1627: \begin{eqnarray}
1628: N_{{\rm BSS},i} & = \; N_{{\rm BSS,obs},i} & - \quad
1629: N_{{\rm VRS},i}\,{} f_{{\rm BSS}/{\rm VRS}}\\
1630: N_{{\rm RGB},i} & = \; N_{{\rm RGB,obs},i} & - \quad
1631: N_{{\rm VRS},i}\,{} f_{{\rm RGB}/{\rm VRS}},
1632: \end{eqnarray}
1633: where $N_{{\rm BSS,obs},i}$ ($N_{{\rm RGB,obs},i}$) is the number of
1634: BSSs (RGB stars) observed in the annulus.
1635:
1636:
1637: In the case of Ursa Minor, we do not have enough data at large
1638: radial distances from the centre, and therefore cannot obtain a local estimate
1639: for $f_{{\rm RGB}/{\rm VRS}}$ and $f_{{\rm BSS}/{\rm VRS}}$. For this
1640: reason, we use the values obtained for Draco also for Ursa Minor. This
1641: is not optimal, but not unreasonable, as the two dSphs are at comparable
1642: Galactic latitudes.
1643:
1644:
1645:
1646:
1647: \subsection{HB foreground contamination}
1648:
1649: The above procedure for RGB and BSSs could also be used for HB stars.
1650: However, for HB stars we adopt a more straightforward technique, making use
1651: of the fact that the foreground does depend on colour, whereas it is
1652: nearly independent of magnitude (at least in the range considered in our CMD).
1653:
1654: Such a fact cannot be exploited in the case of RGB and BSSs, because it
1655: requires a CMD region which is both in the same colour range as BSSs or
1656: RGB, and is populated by foreground stars only. However, for HB stars, the
1657: dashed regions labelled by 4 in Fig.~\ref{fig:cmd_draco_full} are at exactly the
1658: same ($V-I$) range of the regions (labelled 3) where HB stars are selected, and
1659: are almost exclusively populated by foreground stars.
1660:
1661:
1662: In order to account for the different foreground level for RHB and BHB stars,
1663: we further divided region 4 of Fig.~\ref{fig:cmd_draco_full} into two sub-regions:
1664: a blue one with the same colour range of BHB, and a red one with the same
1665: colours of RHB. We will refer to stars in the two sub-regions as to the
1666: fgBHB, and the fgRHB stars, respectively.
1667:
1668: Then, the corrected numbers of RHB (BHB) stars in the $i$-th annulus are
1669: \begin{eqnarray}
1670: N_{{\rm RHB},i} & = \; N_{{\rm RHB,oss},i} -
1671: N_{{\rm fgRHB},i} & {A_{\rm RHB}\over A_{\rm fgRHB}}\\
1672: N_{{\rm BHB},i} & = \; N_{{\rm BHB,oss},i} -
1673: N_{{\rm fgBHB},i} & {A_{\rm BHB}\over A_{\rm fgBHB}},
1674: \end{eqnarray}
1675: where $N_{{\rm RHB,oss},i}$($N_{{\rm BHB,oss},i}$) is the number of
1676: RHB (BHB) stars which was actually observed in the annulus,
1677: $N_{{\rm fgRHB},i}$($N_{{\rm fgBHB},i}$) is the number of fgRHB (fgBHB) stars
1678: in the annulus, and $A_{\rm RHB}/A_{\rm fgRHB}$
1679: ($A_{\rm BHB}/A_{\rm fgBHB}$) is a correction factor which accounts for
1680: the different extensions of the various regions in the CMD.
1681:
1682: Foreground subtraction is then carried out by subtracting the number of
1683: fgBHB (fgRHB) stars in the annulus (after a correction accounting for the
1684: ratios of the CMD areas) from the number of BHB (RHB) stars in the annulus.
1685:
1686: We note that this method of foreground subtraction has a slight
1687: dependence on the radial distance, the foreground level within the core
1688: radius being generally higher than outside. This is because, in addition
1689: to subtracting Milky Way stars in the foreground component, this technique also
1690: accounts for extra-effects (like binaries, blending, errors in the
1691: observed magnitude and contamination from other stellar types, like
1692: RGB).
1693:
1694: As a sanity check, we compared the distribution of HB stars obtained by using
1695: this method of foreground subtraction and that obtained using the same
1696: procedure as for RGB and BSSs. The differences in the inner annuli are
1697: negligible. In the outer annuli ($>2\,{}r_c$, where $r_c$ is the core
1698: radius) the difference is larger, but remains within the (quite large)
1699: Poissonian error bars.
1700:
1701:
1702:
1703: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1704:
1705: \section{A test of the young star hypothesis through isochrones}
1706:
1707: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% FIGURE B1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1708: \begin{figure*}
1709: \center{{
1710: \epsfig{figure=figb1a.eps,height=8cm}
1711: \epsfig{figure=figb1b.eps,height=8cm}
1712: }}
1713: \caption{\label{fig:figb1} Reddening and distance corrected isochrones
1714: of single stellar populations superimposed to the CMD of the central
1715: region of Draco (left panel; here, in order to reduce foreground
1716: contamination, we plot only stars with $r\leq28.3'$) and Ursa Minor
1717: (right). In both cases, the left line refers to a metallicity
1718: [Fe/H]=-2.3, the middle one to [Fe/H]=-2.0, and the right one to
1719: [Fe/H]=-1.5. We assumed ages of 2.0 and 2.5 Gyr in the case of Draco and
1720: Ursa Minor, respectively. Boxes are the same as in Fig.~\ref{fig:fig1}.}
1721: \end{figure*}
1722: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1723:
1724: Although we have shown that the properties of the observed BSS
1725: population are fully compatible with the expectations for `real' BSSs,
1726: the `young star' interpretation provided by A01 for Draco still
1727: remains viable.
1728:
1729: A more direct test of the nature of this population can be performed by
1730: looking for other hints of a relatively young population. Actually, the
1731: A01 interpretation of the BSSs in Draco was based on the
1732: observation of a small concentration of stars in a region of their CMD
1733: (the `red clump', i.e. region 13 of their fig. 13) which should not
1734: be populated if no star formation occurred in the last 10 Gyr.
1735:
1736: We performed a similar test by means of the isochrones of the Padova
1737: group (see Girardi et al. 2002 ; see also
1738: {\tt http://stev.oapd.inaf.it/$\sim{}$lgirardi/cmd}).
1739:
1740: We plotted a set of theoretical isochrones over the Draco and Ursa Minor
1741: CMDs\footnote{We assumed a reddening E($B-V$)=0.03 for both galaxies, a
1742: value for which a vast consensus exists. The distance modulus of Ursa
1743: Minor was chosen to be 19.41 [Bellazzini et al. (2002) and C02 found
1744: 19.41$\pm$0.12 and 19.40$\pm$0.10, respectively]. The distance
1745: modulus of Draco is more controversial, as recent determinations yielded
1746: relatively different values: A01, and Bonanos et al. (2004) found
1747: compatible values (19.5$\pm$0.2 and 19.40$\pm$0.15$\pm$0.02,
1748: respectively); but Bellazzini et al. (2002) found a significantly higher
1749: value (19.84$\pm$0.14). We adopted the intermediate value of 19.60. We cannot make an estimate from our data, as the tip of the RGB in our WFC data is beyond the saturation limit.},
1750: % is you include the -9 category by the way you will populate the upper part
1751: % of the CMDs. In forming the band-merge cataloge a correction is made for
1752: % saturation which extends the dynamic range by about 1-2 mags.
1753: varying both the age and the metallicity of the stellar population (see
1754: Fig. B1 for some examples). From such isochrones it is clear that BSSs
1755: lie close to the isochrones describing low-metallicity
1756: ($[Fe/H]\leq-1.5$, perfectly compatible with current estimates for Draco
1757: and Ursa Minor) stellar populations with ages between 2 and 3 Gyr.
1758:
1759: For Draco we chose to combine a Chabrier (2003) log-normal IMF with an
1760: isochrone for an age of 2 Gyr and a metallicity [Fe/H]=-2.0, in order to
1761: estimate the number of stars which should be expected in other regions
1762: of the CMD if all of the observed BSSs are actually part of an
1763: intermediate-age population. Within this scenario, the regions where we
1764: expect the maximum number of intermediate-age stars and the minimum
1765: contamination from the old population, are:\begin{enumerate}
1766: \item{}{the young MS just below the BSS selection box;}
1767: \item{}{the bright and faint part of of the BSS selection box;}
1768: \item{}{the red clump.}
1769: \end{enumerate}
1770: In Table~B1 we list the theoretical predictions from the
1771: isochrones ($N_{\rm pred}$), the total number of stars observed in each
1772: CMD region ($N_{\rm raw}$), the estimated foreground contamination
1773: ($N_{\rm fg}$), and the number of observed stars after the foreground
1774: subtraction ($N_{\rm obs}$).
1775:
1776: %% It is clear that the predictions from the young star hypothesis are
1777: %% quite compatible with our observations of Draco: the ratio of faint BSSs
1778: %% to bright BSSs is slightly lower than expected, but the difference is
1779: %% just at the $1\sigma$ level; on the other hand, the predicted number of
1780: %% young MS and red clump stars is perfectly compatible with observations,
1781: %% as both these CMD regions are likely to be contaminated by the old Draco
1782: %% stellar population (old MS stars close to the turn-off for the young MS
1783: %% region, HB and RGB stars for the red clump region).
1784:
1785:
1786: It is clear that the predictions from the young star hypothesis are
1787: quite compatible with our observations of Draco: the ratio of faint BSSs
1788: to bright BSSs is slightly lower than expected ($1.75\pm0.49$ instead of
1789: $2.29$), but the difference is just at the $1-\sigma$ level; on the
1790: other hand, the predicted number of young MS and red clump stars is
1791: perfectly compatible with observations, as both these CMD regions are
1792: likely to be contaminated by the old Draco stellar population (old MS
1793: stars close to the turn-off for the young MS region, RGB and especially
1794: - given the partial superposition of the two regions - HB stars for the
1795: red clump region).
1796:
1797: %% We applied the same method also to Ursa Minor, using an isochrone age
1798: %% of 2.5 Gyr. Results are summarized in Table B2, where we omitted the
1799: %% Young MS region because of the very strong contamination from the old
1800: %% MS. The main result is that the number of stars in the red clump region
1801: %% is significantly lower than expected from the young stars hypothesis:
1802: %% the predicted 9.4 red clump stars are within the $1\sigma$ upper limit
1803: %% from the observations (10.4), but contamination from HB and RGB stars is
1804: %% surely present, and there remains very little space for the young star
1805: %% hypothesis. However, we note that a distance modulus 0.1-0.2
1806: %% magnitude larger than what we assumed could ``hide'' the red clump stars
1807: %% within the HB.
1808:
1809: We applied the same method also to Ursa Minor, using an isochrone age of
1810: 2.5 Gyr. Results are summarized in Table B2, where we omitted the Young
1811: MS region because of the very strong contamination from the old MS. The
1812: number of stars in the red clump region appears to be extremely close to
1813: the prediction from the young stars hypothesis, but a strong
1814: contamination is surely present, as the red clump selection box
1815: ($19.62\ge V\ge 19.12$, $0.90\ge (V-I) \ge 0.50$) largely superimposes
1816: with the HB (see Fig. B1). Such a strong contamination accounts for most
1817: of the `excess' stars in the considered CMD region. However, the young
1818: star hypothesis might still be viable, because most of the red clump
1819: stars might be `hidden' within the HB.
1820:
1821: In summary, the interpretation that BSS candidates in Draco are
1822: intermediate-age stars can neither be ruled out, nor be confirmed by the
1823: isochrone method applied to our observations. Only a spectral analysis
1824: of stars in the red clump region could solve the uncertainty. In Ursa
1825: Minor such an interpretation is hardly compatible with current data (see
1826: also C02), but cannot be completely ruled out.
1827:
1828: We point out that in both galaxies the mass of the intermediate-age
1829: population needed to explain the BSS candidates is just about
1830: $10^4\Msun$, which is a very small fraction of the mass of Draco or
1831: Ursa Minor. If the age spread is $\gtrsim$ 1 Gyr, the implied star
1832: formation rate is $\lesssim{}10^{-5}\,\Msun\,{}{\rm yr}^{-1}$, comparable
1833: to the estimates shown in figs. 14 and 16 of A01, and much lower
1834: than any observed star formation rate in dwarf galaxies.
1835:
1836: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% TABLE B1%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1837: \begin{table}
1838: \label{isoc_table_draco}
1839: \begin{center}
1840: \caption{Comparison of isochrone predictions (age 2.0 Gyr, [Fe/H]=-2.0)
1841: with observations for Draco} \leavevmode
1842: \begin{tabular}[!h]{lcccc}
1843: \hline
1844: CMD region
1845: & $N_{\rm pred}$
1846: & $N_{\rm raw}$
1847: & $N_{\rm fg}$
1848: & $N_{\rm obs}$\\
1849: \hline
1850: Young MS & 36.8$\pm$3.4 & 43 & 2.3 & 40.7$\pm$7\\
1851: BSS faint & 84.2$\pm$7.7 & 76 & 2.3 & 73.7$\pm$9\\
1852: BSS bright & 36.8$\pm$3.4 & 45 & 3.0 & 42.0$\pm$7\\
1853: Red clump & 15.3$\pm$1.4 & 101 & 53.3 & 47.7$\pm$11\\
1854: \noalign{\vspace{0.1cm}}
1855: \hline
1856: \end{tabular}
1857: \end{center}
1858: \end{table}
1859: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1860: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% TABLE B2%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1861: \begin{table}
1862: \label{isoc_table_ursa}
1863: \begin{center}
1864: \caption{Comparison of isochrone predictions (age 2.5 Gyr, [Fe/H]=-2.0)
1865: with observations for Ursa Minor} \leavevmode
1866: \begin{tabular}[!h]{lcccc}
1867: \hline
1868: CMD region
1869: & $N_{\rm pred}$
1870: & $N_{\rm raw}$
1871: & $N_{\rm fg}$
1872: & $N_{\rm obs}$\\
1873: \hline
1874: BSS faint & 63.6$\pm$6.4 & 71 & 1.0 & 70.0$\pm$9\\
1875: BSS bright & 36.3$\pm$3.7 & 29 & 1.0 & 28.0$\pm$6\\
1876: Red clump & 11.4$\pm$1.2 & 28 & 16.2 & 11.8$\pm$6\\
1877: \noalign{\vspace{0.1cm}}
1878: \hline
1879: \end{tabular}
1880: \end{center}
1881: \end{table}
1882: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1883:
1884: Finally, the isochrones can also be used to give an indicative estimate
1885: of the upper/lower limit mass of BSSs, which are used to set up our
1886: simulations (see Section 4). For Draco we find that their masses should
1887: be in the range $1.11-1.35\Msun$, whereas for Ursa Minor this range
1888: moves slightly to $1.09-1.34\Msun$.
1889:
1890:
1891: \end{document}
1892: