1: \documentclass[apj,twocolumn]{emulateapj}
2: %\documentclass[12pt,preprint]{aastex}
3:
4: %\usepackage{amsmath}
5: %\usepackage{txfonts}
6:
7: \renewcommand\email\texttt
8:
9: \shorttitle{The discovery of two globular clusters}
10: \shortauthors{Koposov et al.}
11: \slugcomment{Accepted for publication in ApJ}
12: \begin{document}
13:
14: \title{The discovery of two extremely low luminosity Milky Way
15: globular clusters}
16:
17: \author{S. Koposov\altaffilmark{1,2}, J.T.A. de Jong\altaffilmark{1},
18: V. Belokurov\altaffilmark{2}, H.-W. Rix\altaffilmark{1},
19: D.B. Zucker\altaffilmark{2}, N.W. Evans\altaffilmark{2},
20: G. Gilmore\altaffilmark{2}, M.J. Irwin\altaffilmark{2},
21: E.F. Bell\altaffilmark{1}}
22:
23: \altaffiltext{1}{Max Planck Institute for Astronomy, K\"onigstuhl 17,
24: 69117, Heidelberg, Germany;\email{koposov,dejong@mpia.de}}
25: \altaffiltext{2}{Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge CB3 0HA, UK}
26:
27:
28: \begin{abstract}
29: We report the discovery of two extremely low luminosity globular
30: clusters in the Milky Way Halo. These objects were detected in the Sloan
31: Digital Sky Survey Data Release 5 and confirmed with deeper imaging at the Calar
32: Alto Observatory. The clusters, Koposov 1 and Koposov 2, are located at
33: $\sim 40-50$ kpc and appear to have old stellar populations and luminosities of
34: only $M_V \sim -1 mag$. Their observed sizes of $\sim 3$ pc are well within the
35: expected tidal limit of $\sim$10 pc at that distance. Together with Palomar 1,
36: AM 4 and Whiting 1, these new clusters are the lowest luminosity globulars
37: orbiting the Milky Way, with Koposov 2 the most extreme. Koposov 1 appears to
38: lie close to distant branch of the Sagittarius stream. The half-mass relaxation
39: times of Koposov 1 and 2 are only $\sim 70$ and $\sim 55$ Myr respectively
40: (2 orders of magnitude shorter than the age of the stellar populations), so it
41: would seem that they have undergone drastic mass segregation. Since they do not
42: appear to be very concentrated, their evaporation timescales may be as low as
43: $\sim 0.1 t_{\rm Hubble}$. These discoveries show that the structural parameter
44: space of globular clusters in the Milky Way halo is not yet fully
45: explored. They also add, through their short remaining survival times,
46: significant direct evidence for a once much larger population of
47: globular clusters.
48: \end{abstract}
49:
50: \keywords{Galaxy:halo -- Galaxy:globular clusters}
51:
52: \section{Introduction}
53:
54: The population of globular clusters around the Milky Way has been
55: studied extensively and the current census finds the majority at low
56: latitudes in the inner Galaxy ($R_{\rm GC} < 20$ kpc). Globular
57: clusters are almost universally ``old'' [$t_{\rm age} \approx 0.5 -- 1
58: \times t_{\rm Hubble}$], show no convincing evidence for dark matter,
59: and have characteristic luminosities of $10^5L_\odot$ $M_V \sim -8$)
60: and typical sizes of 3 pc. Yet, the observed range of structural
61: properties (e.g. mass, size, and concentration) is quite wide. This
62: range is of great interest, as it appears to be determined by a set of
63: astrophysical processes: the initial structure and orbit; subsequent
64: external processes, such as galactic tides and dynamical friction; and
65: ensuing mass segregation, evaporation, and core collapse~\citep[see,
66: e.g.,][]{gnedin_ostriker,meylan97}. Indeed, there has long been a
67: sense that the observed population of Galactic globular clusters
68: mainly reflects the subset of objects that could survive for $\sim
69: t_{\rm Hubble}$. In individual cases, there is clear evidence for
70: internal reshaping processes~\citep[as in \object{M15},][]{sosin97} and tidal
71: disruption~\citep[as in \object{Pal5},][]{odenkirchen}. Within this context,
72: identification and study of globular clusters with extreme properties is undoubtely of
73: great interest.
74:
75: Our census of objects at the outskirts of the Milky Way has increased
76: rapidly in the last few years, mostly based on large-area CCD surveys
77: such as the Sloan Digital Sky Survey~\citep[SDSS;][]{york}. Recent
78: searches for Galactic halo objects have not only found many dwarf
79: galaxies~\citep{willman,canven,5pack,irwin}, but also added two faint
80: and extended objects that may be Milky Way globular clusters. The newcomers,
81: \object{Willman 1} and \object{Segue 1}, both have distorted
82: irregular isopleths, perhaps indicating ongoing tidal disruption.
83: \object{Willman 1} seems to show some evidence for dark matter and metallicity
84: spread \citep{martin}, casting some doubt on whether it is a
85: globular cluster at all.
86:
87: Here, we announce the discovery of two new, distant, extremely faint
88: and compact ($\sim 3$ pc) globular clusters, named \object{Koposov 1} and
89: \object{Koposov 2}, first detected in SDSS Data Release 5 (DR5) and
90: subsequently confirmed with deeper imaging at Calar Alto. The total
91: luminosity of Koposov 2 appears to be $\sim\ -1$ mag, lower than that of
92: the faintest Galactic globular known to date, \object{AM 4}
93: \citep[$-1.4$ mag,][]{am4}. Koposov 1 is not much brighter; at
94: $M_{V,tot} \sim -2$ mag, it has the third-lowest luminosity. In total,
95: only 3 out of the previously known $\sim 160$ Galactic clusters, have
96: comparably low luminosity and small sizes: AM 4, \object{Palomar 1}, and
97: \object{Whiting 1} \citep{whiting}. Willman 1 and Segue 1 also have extremely
98: low luminosities, but are an order of magnitude larger.
99:
100: In this paper, we describe the deep follow-up data confirming the
101: discoveries and give estimates of the structural parameters of the new objects (see Table~\ref{prop_table}).
102: We argue that the discovery of these two low-mass globulars in less than 1/5 of
103: the sky may mean that a substantial population of such clusters lurks in
104: the outer halo of the Milky Way.
105:
106:
107: \section{Discovery and observations}
108:
109: \begin{figure*}
110: \plottwo{f1__.eps}{f2__.eps}
111: \caption{$3\arcmin \times 3\arcmin$ SDSS cutout images
112: of Koposov 1 and 2. The bright star in the
113: center of Koposov 1 is a foreground star with $V\sim14.5^m$ and large
114: proper motion $(\mu_\alpha,\mu_\delta )\sim(-32,-12 ) \mu as\,yr^{-1}$,
115: according to the USNO-B1 catalog~\citep{usno}. The bright extended
116: object near the center of Koposov 2 is a background galaxy.}
117: \label{sdss_cutouts}
118: \end{figure*}
119:
120: \begin{deluxetable}{lcc}
121: \tablecaption{Properties of Koposov 1 and Koposov 2}
122: \tablecolumns{2}
123: %\tablehead{\colhead{Parameter}}
124: \tablehead{\colhead{Parameter}&\colhead{Koposov 1}&\colhead{Koposov 2}}
125: \startdata
126: Coordinates (J2000) & 11:59:18.5 +12:15:36 & 07:58:17.00 +26:15:18 \\
127: Coordinates ($\ell,b$) & $(260.98\degr, 70.75\degr)$
128: & $(195.11\degr, 25.55\degr)$\\
129: Distance & $\sim 50$ kpc & $\sim 40$ kpc\\
130: Size & $\sim 3$ pc & $\sim 3$ pc\\
131: $M_V$ & $\sim -2^m$ & $\sim -1^m$ \\
132: Relaxation Time & $\sim 70$ Myrs & $\sim 55$ Myrs \\
133: Tidal radius & $\sim 11$ pc & $\sim 9$ pc\\
134: \enddata
135: \label{prop_table}
136: \end{deluxetable}
137:
138: The two new globular clusters were originally selected among other
139: candidates in the course of our systematic search for small-scale substructure
140: in
141: the Milky Way halo. The aim of the search was to detect all
142: significant small-scale stellar overdensities above the slowly varying
143: Galactic background that are likely to be either dwarf spheroidal
144: galaxies or globular clusters. A detailed description of the algorithm
145: and its efficiency will be provided in a future paper, and we only
146: present here a brief outline of the method. The algorithm is based on
147: the so-called Difference of Gaussians method, first developed in
148: Computer Vision~\citep{babaud,lindenberg}. Starting from a
149: flux-limited catalog of stellar positions, the number-counts map in
150: $(\alpha,\delta)$ plane is convolved with a filter optimized for the
151: detection of overdensities, namely the difference of two two-dimensional Gaussians
152: \citep{koposov}. Having zero integral, the kernel guarantees that the
153: convolution with a constant (or slowly varying) background will result
154: in zero signal. When the data contain an overdensity with a size
155: comparable to the size of the inner Gaussian, the filter will be close
156: to optimal.
157:
158: We applied this filtering procedure to the entire stellar subset of
159: the DR5 source catalog with $r < 22^m, g-r < 1.2^m$. In our analysis we used
160: the photometry cleaned by switching on quality flags as described in SDSS SQL
161: pages~\footnote{\url{http://cas.sdss.org/dr5/en/help/docs/realquery.asp\#flags}}
162: This minimizes the influence of various artefacts including those caused by
163: proximity
164: of very bright or extended objects. In the resulting map that had been convolved
165: with a 2\arcmin\ kernel, we found two very compact objects among other
166: overdensities ranked highly according to their statistical significance.
167: Figure~\ref{sdss_cutouts} shows the SDSS images, and Figure~\ref{sdss_dots}
168: shows the spatial distribution of extracted sources, where central
169: concentrations of stars are clearly visible.
170: These concentrations are detected at high level of significance. The areas of
171: 1\arcmin\ radius marked by circles centered on Koposov 1 and 2 plotted in
172: Figure~\ref{sdss_dots} contain 22 objects and 23 objects, respectively, while
173: mean density of g-r$<0.6$, r$>20 mag$ stars should produce approximately 2.5
174: objects, which
175: implies a high statistical significance of the overdensities; for pure
176: Poisson distribution of objects, the probability to find such group of stars in
177: all DR5 is around $10^{-9}$.
178:
179:
180:
181: \begin{figure*}
182: \plotone{f3.eps}
183: \caption{The spatial distribution of the objects in the area of Koposov 1
184: and Koposov 2. All objects classified as stars with colors (g-r)$<0.6^m$ and
185: r$>20^m$ in the area $0.3\degr\times0.3\degr $ are shown. The circles with
186: 1\arcmin\ radii centered on the objects are overplotted.
187: }
188: \label{sdss_dots}
189: \end{figure*}
190:
191: \begin{figure*}
192: \plotone{f4.eps}
193: \caption{The residual $g-r$ versus $g$ Hess diagrams of the clusters from
194: the SDSS data. In each case the residual Hess diagram is costructed by
195: subtracting the normalized background Hess diagram from the Hess diagram of
196: stars lying within 2\arcmin.5 radius from the centers of objects}
197: \label{sdss_hess}
198: \end{figure*}
199:
200: The differential Hess diagrams for stars within $2\arcmin.5$ radius centered on
201: the objects are shown in Figure~\ref{sdss_hess}. There is a clear excess of blue
202: stars ($g-r < 0.5$), which we interpret as main-sequence turnoff stars at
203: $r\sim 22$, which roughly corresponds to distances of $\sim 50$kpc.
204:
205: To confirm the nature of discovered candidates and quantify their structural
206: and population properties, we acquired follow-up GTO observations in 2007 January
207: on the 2.2m telescope at Calar Alto using the CAFOS camera. This camera has
208: a $2k\times2k$ CCD with a $16\arcmin\times16\arcmin$ field of view and a pixel
209: scale of $0\arcsec.5\,pixel{^-1}$. We observed each object for a total of 2 hr in
210: Johnson $B$ and 1.5 hr in Cousins $R$. The integrations were split into five
211: individual dithered exposures for cosmic ray and bad pixel
212: rejection. The observations were carried out in good photometric
213: conditions with a seeing of $1\arcsec-1.3\arcsec$. The data were
214: bias-subtracted and flat-fielded. The individual frames were
215: WCS-aligned, drizzled, and median-combined using our software and the
216: SCAMP and SWARP programs~\citep{scamp}. The combined $B$ band images of
217: the objects are shown in Figure~\ref{ca_images}.
218:
219: \begin{figure*}
220: \plotone{f5_.eps}
221: \caption{$B$ band Calar Alto view of Koposov 1 and Koposov 2. The
222: $2\arcmin \times 2\arcmin$ images are centered on the clusters
223: (north is up, east is left).}
224: \label{ca_images}
225: \end{figure*}
226:
227:
228: \begin{figure*}
229: \plottwo{f6.eps}{f7.eps}
230: \caption{\it Left panel, left half \rm: $B$ vs. $B-R$ CMDs derived from the
231: Calar Alto data for stars lying within 2\arcmin\ of Koposov 1 with 8 Gyr and
232: [Fe/H]=-2~\citet{girardi} isochrones overplotted. \it Left panel, right half\rm
233: : For comparison, the CMDs of stars in the annulus centered on Koposov 1 defined
234: by radii 3.2\arcmin\ and 3.7\arcmin are plotted. \it Right panel, left half\rm : $B$ vs.
235: $B-R$ CMDs of stars lying within
236: 1.\arcmin2 of Koposov 2 with 8 Gyr and [Fe/H]=-2~\citet{girardi} isochrones
237: overplotted.\it Right panel, right half\rm : for comparison, the CMDs of stars
238: in the annulus centered on
239: Koposov 2 defined by radii 2\arcmin\ and 2.3\arcmin are plotted.}
240: \label{cmds}
241: \end{figure*}
242:
243:
244:
245: The central stellar overdensities are clearly corroborated by the
246: Calar Alto photometry, which is nearly 2 mag deeper than the
247: original SDSS data. While the follow-up data are quite deep, the stars
248: are subject to significant crowding, due to the compactness of the
249: clusters. Therefore, for the purposes of robust source detection and
250: photometry, we used the DAOPHOT/ALLSTAR software \citep{daophot}. To
251: get the absolute calibration of the photometry from each frame, we
252: cross-matched the DAOPHOT sources with the SDSS catalog using the
253: Virtual Observatory resource SAI
254: CAS~\footnote{\url{http://vo.astronet.ru/cas} } \citep{sai_cas}. To
255: convert the Sloan $g$ and $r$ magnitudes into the Johnson-Cousins
256: photometric system, we used the conversion coefficients
257: from~\citet{smith}. The resulting $B$ versus
258: $B-R$ color-magnitude diagrams (CMDs) of the central regions of the objects together
259: with the CMDs of the comparison fields, extending to $B\sim
260: 23.5^m-24^m$, are shown in the Figure~\ref{cmds}. The median photometric
261: accuracy of the data is 0.05-0.1 mag. The CMDs clearly show the presence of the main sequences near the centers of the
262: objects, while they are absent in the the comparison fields. The
263: statistical significance of the overdensities is also clearly supported by the new
264: data. The CMD of objects within 2\arcmin\ from the center of Koposov 1
265: contain 96 objects, while the background density inferred from the comparison
266: field should give around 23 objects, which gives a 15 $\sigma$ deviation. For
267: Koposov~2 , the number of objects within 1.2\arcmin\ is 92, while the background
268: density from the comparison field should produce around 24 objects, which gives
269: a 14~sigma deviation. In the next section we will discuss the properties of the
270: objects which can be derived from the follow-up data.
271:
272: \section{Properties}
273: \label{properties_section}
274: The CMDs of the objects from the Calar
275: Alto data (Fig.~\ref{cmds}) clearly show a distribution of stars which can be
276: attributed quite convincingly to an old main sequence. In the case of Koposov 1,
277: the main-sequence turnoff is clear-cut, while for the second cluster it is not
278: so well defined. To estimate the distances to the objects, we
279: overplot in Figure~\ref{cmds} the 8 Gyr [Fe/H]=-2 isochrones from
280: \citet{girardi}. For Koposov 1, this gives a distance of 50 kpc. For Koposov 2,
281: the estimate is 40$\pm$5 kpc, but it is not well constrained due to a lack
282: of main-sequence turnoff stars. The angular diameters of the clusters are
283: $<0.5\arcmin$, which translates into a physical size of $r \sim
284: 5$pc. Unfortunately, the number of stars detected in the central
285: regions is not enough to precisely measure half-light radii of the
286: objects; our best estimate is $r_h \sim 3$pc. For Koposov 1, we
287: subtracted the bright foreground star near the center, integrated the
288: light of the whole cluster in apertures and fitted it to a Plummer
289: profile with $r_h=3$pc. For Koposov 2, we performed a maximum
290: likelihood fit with $r_h \sim 3$pc. Moreover, the minuscule number of
291: stars in both clusters does not allow us to firmly establish their
292: total luminosities. Our estimate of $-1 \ga M_V \ga -2$ is based on the
293: absence of the giants in these clusters and the visible similarity of
294: the CMDs to that of the lowest luminosity globular cluster AM4
295: \citep[$M_V=-1.6$,][]{am4}. We checked that estimate by a simple Monte Carlo
296: experiment: using the Salpeter IMF and Girardi isochrones we simulated fake
297: clusters and deduced that the clusters with $-1\ga M_V \ga -2 $ have a number
298: of stars within 1.5-2 mag below the turnoff is close to the
299: observed number of stars (50-70) in our objects. We must say also that due to
300: the intrinsic faintness of the clusters and low number of stars in them the
301: estimates of the total luminosity and especially the age have large
302: uncertainties. However, with the existing data we cannot do any better. Much deeper
303: and more accurate photometry may be required to get precise age/luminosity
304: measures. The spectroscopic observations would be interesting in constraining
305: the metallicity of these objects, which is currently completely unknown.
306:
307: We note that the CMD of Koposov 1 shows
308: several stars brighter and bluer than the tentative main-sequence turnoff, which
309: we interpret as blue stragglers. This hypothesis is not implausible
310: considering the low luminosity of the cluster and taking into account
311: the observed anti-correlation between the frequency of blue stragglers
312: and the luminosity of the globular cluster \citep{piotto}.
313:
314: The distance and the position of Koposov 1 suggest that this cluster
315: may be related to the Sagittarius tidal stream. Its location is a good
316: match to the distant tidal arm discovered in
317: \citet{field_of_streams}. Figure~\ref{sagi} shows the arms of the
318: Sagittarius stream in the DR5 slice around $\delta \sim 10\degr$ and
319: the position of Koposov 1.
320:
321: \section{Discussion}
322:
323: \begin{figure}
324: \plotone{f8.eps}
325: \caption{Right ascension vs. distance for the A and C branches of
326: the Sagittarius stream~\citep[see][]{field_of_streams}. The
327: position of Koposov 1 is marked by a star.}
328: \label{sagi}
329: \end{figure}
330:
331:
332:
333: Figure~\ref{size_mag} shows Koposov 1 and 2 on the size-luminosity
334: plane along with other Galactic globular clusters. This illustrates
335: how unusual Koposov 1 and 2 are in their structural properties. It
336: appears that the detection of these clusters contributes to growing
337: evidence for a large population of small and extremely faint objects
338: (including Palomar 1, AM 4, E3 and Whiting 1). There is a
339: clear indication as well that this sub-population of globular clusters may
340: have significantly younger ages than classical globular clusters:
341: Palomar 1~\citep{sarajedini} and Whiting 1~\citep{whiting_new} have ages
342: between 4 and 6 Gyrs. The current estimate for the age of Koposov 1 is
343: $\approx 8$ Gyr, and the age of E3 globular cluster is $\approx$ 10 Gyr. This group
344: of clusters is also quite apparent on the galactocentric distance
345: versus luminosity plane shown in Figure~\ref{dist_mag}. At least 2
346: out of these 5 unusual clusters (Whiting 1 and Koposov 1) seem to be
347: associated with the Sagittarius dwarf galaxy.
348: Two quantities that are crucial for the long term evolution and
349: survival of Koposov 1 and 2 are the relaxation time and the expected
350: tidal radius. For the half-mass relaxation time, we find
351: using equation (2-63) of \citet{spitzer} or equation (72) of \citet{meylan97},
352: %
353: $$t_{\rm rh} = 0.14 \frac{M_{\rm tot}^{1/2} {R_{\rm hl}}^{3/2}}{\langle m_*\rangle G^{1/2}
354: \ln(\Lambda)}=70\ {\rm and}\ 55\ {\rm Myr}$$
355: %
356: respectively for Koposov 1 and Koposov 2.
357: Here, we have assumed $L\approx 200 L_\odot$, $M/L \approx 1.5$,
358: $\langle m_* \rangle \approx 0.6 M_\odot$ and N=500 for Koposov 2,
359: while for Koposov 1, we have assumed twice as many stars, using the
360: observational estimates of \S~\ref{properties_section}. This means that
361: both clusters have extremely short relaxation times, less than 1\% of
362: $t_{\rm Hubble}$ and $t_{rh} \approx 0.01 t_{age,*}$. The most immediate effect
363: of two-body relaxation is mass segregation, which should be quite drastic given
364: the apparent stellar population age. The
365: expected tidal truncation of these clusters occurs at~\citep[see,
366: e.g.,][]{innanen}
367: %
368: $$r_t = 0.43\left(\frac{M_{\rm cluster}}{M_{\rm MW}}\right)^{1/3}
369: \times R_{\rm peri} = 11\ {\rm and}\ 9\ {\rm pc}$$
370: %
371: where we have assumed an orbital eccentricity of 0.5, and that the
372: clusters are now near apocenter (hence $R_{peri} \approx 16$ kpc), a
373: Milky Way circular speed of 190 $km\,s^{-1}$ at 16 kpc and a cluster (stellar)
374: mass of 600 and 300 $M_{\odot}$ for Koposov 1 and 2, respectively. Hence, the
375: detectable extent of the globular clusters (3
376: pc) falls well within the tidal limit. From this argument, the
377: clusters are under no threat of destruction by tidal forces. Although
378: formal profile fits are not feasible with so few stars, the stellar
379: distributions (see Figs.~\ref{sdss_cutouts} and \ref{ca_images}) are
380: well localized, but not centrally concentrated by globular cluster
381: standards; a core to tidal radius ratio of the observed stellar
382: distribution of 4 seems reasonable, implying a concentration parameter
383: of $c \equiv \log (r_t/r_c) \approx 0.5$. For such low concentrations,
384: the evaporation timescale $t_{\rm ev}$, which is the time-scale over
385: which two-body relaxation drives stars to beyond the escape velocity,
386: is $t_{\rm ev} \approx 1.5 t_{\rm cc} \approx 12 t_{\rm
387: rh}$(where $t_{\rm cc}$ is core collapse time)~\citep[Figure 17 and 19 in
388: ][]{gnedin_lee_ostriker}For Koposov 1 and 2, this implies
389: evaporation time-scales of 0.7 and 1.1 Gyr, respectively. This estimate
390: of $t_{ev} \sim 0.1 t_{Hubble}$ may be an underestimate, if the brightest stars
391: which we observe are more concentrated than the faint stars due to mass
392: segregation; then the total mass and half-mass radius can be larger.
393: Nonetheless, this estimate makes it clear that the present structural and
394: dynamical state cannot have prevailed, even approximately, for a time-span of
395: $\sim$ 10 Gyr.
396: The above arguments hold irrespective of whether Koposov 1 and Koposov
397: 2 were once part of a satellite galaxy, because they are mostly
398: derived from internal evolution factors. This discrepancy of
399: time-scales is more pronounced in Koposov 1 and 2, because their
400: relaxation time-scales are shorter than those of Palomar 1 and Whiting 1, which
401: in any case have accurate photometry suggesting younger ages of $\sim
402: 4-6$\,Gyr.
403:
404: \begin{figure}
405: \plotone{f9.eps}
406: \caption{Size vs. absolute magnitude plot for Galactic globular
407: clusters. The data from the \citet{harris} catalog are plotted with
408: diamonds. Squares mark the locations of the recently discovered
409: globular clusters Willman 1, Segue 1 and Whiting 1. Koposov 1 and 2
410: are shown as stars.}
411: \label{size_mag}
412: \end{figure}
413:
414:
415:
416: \begin{figure}
417: \plotone{f10.eps}
418: \caption{Galactocentric distance vs. magnitude plot for
419: Galactic globular clusters. Symbols are as in
420: Fig.~\ref{size_mag}.}
421: \label{dist_mag}
422: \end{figure}
423:
424:
425: At face value, Koposov 1 and 2 have survival times in their current
426: state of $\sim 0.1 t_{\rm Hubble}$, and were found in a search of 20\%
427: of the whole sky (SDSS DR5). The naive multiplication of these factors
428: points to a large parent population of $\sim 100$ objects. The most
429: likely reservoir for this parent population is the globular clusters,
430: and possibly even old open clusters, in satellite galaxies that have
431: been accreted, like the Sagittarius. In objects like Koposov 1 and 2,
432: it is clear that the very short relaxation and evaporation times must
433: lead to drastic mass segregation and the expulsion of basically all
434: low-mass stars (this line of reasoning lead us to the modest $M/L
435: \approx 1.5$). This gives new life to the view that truly
436: \textit{many} of the accreted globular clusters must have been
437: destroyed. Yet, it is also clear that the actual dynamical prehistory
438: and future of these clusters requires much more careful modelling.
439: The small number of stars makes them ideal subjects of direct \textit{N}-body
440: calculations. But regardless of their dynamical evolution, these
441: clusters manifestly demonstrate that the parameter space of globular
442: clusters in the Milky Way is not yet fully explored.
443:
444:
445: \acknowledgements{
446: Based on observations collected at the Centro Astron\'omico Hispano
447: AlemÃ\'an (CAHA) at Calar Alto, operated jointly by the Max-Planck
448: Institut f\"ur Astronomie and the Instituto de Astrof\'isica de
449: Andaluc\'ia (CSIC). Funding for the SDSS and SDSS-II has been provided
450: by the Alfred P. Sloan Foundation, the Participating Institutions,
451: the National Science Foundation, the U.S. Department of Energy, the
452: National Aeronautics and Space Administration, the Japanese
453: Monbukagakusho, the Max Planck Society, and the Higher Education
454: Funding Council for England. The SDSS Web Site is
455: http://www.sdss.org/. The SDSS is managed by the Astrophysical
456: Research Consortium for the Participating Institutions. The
457: Participating Institutions are the American Museum of Natural
458: History, Astrophysical Institute Potsdam, University of Basel,
459: Cambridge University, Case Western Reserve University, University of
460: Chicago, Drexel University, Fermilab, the Institute for Advanced
461: Study, the Japan Participation Group, Johns Hopkins University, the
462: Joint Institute for Nuclear Astrophysics, the Kavli Institute for
463: Particle Astrophysics and Cosmology, the Korean Scientist Group, the
464: Chinese Academy of Sciences (LAMOST), Los Alamos National
465: Laboratory, the Max-Planck-Institute for Astronomy (MPIA), the
466: Max-Planck-Institute for Astrophysics (MPA), NewMexico State
467: University, Ohio State University, University of Pittsburgh,
468: University of Portsmouth, Princeton University, the United States
469: Naval Observatory, and the University of Washington. This research
470: has made use of the SAI Catalog Access Services, Sternberg
471: Astronomical Institute, Moscow, Russia. S. Koposov is supported by
472: the DFG through SFB 439 and by a EARA-EST Marie Curie Visiting
473: fellowship.}
474: \begin{thebibliography}{}
475:
476: %1971
477: %\bibitem[Spitzer \& Hart(1971)]{spitzer} Spitzer, L.~J., \&
478: %Hart, M.~H.\ 1971, \apj, 166, 483
479:
480: %1986
481: \bibitem[Babaud et al.(1986)]{babaud} Babaud J., A. P. Witkin,
482: M. Baudin, and R. O. Duda. 1986, IEEE Trans. Pattern Anal. Mach. Intell.,
483: 8, 1, 26
484:
485: %2006
486: \bibitem[Belokurov et al.(2006)]{field_of_streams} Belokurov, V., et
487: al.\ 2006, \apjl, 642, L137
488:
489: %2007
490: \bibitem[Belokurov et al.(2007)]{5pack} Belokurov, V., et
491: al.\ 2007, \apj, 654, 897
492:
493: \bibitem[Bertin(2006)]{scamp} Bertin, E.\ 2006, ASP
494: Conf.~Ser.~351: Astronomical Data Analysis Software and Systems XV, 351,
495: 112
496:
497: \bibitem[Carraro et al.(2007)]{whiting_new} Carraro G., A\&A,
498: submitted, astro-ph/0702253
499:
500: %2000
501: \bibitem[Girardi et al.(2000)]{girardi} Girardi, L., Bressan,
502: A., Bertelli, G., \& Chiosi, C.\ 2000, \aaps, 141, 371
503:
504: %1997
505: \bibitem[Gnedin et al.(1999)]{gnedin_lee_ostriker} Gnedin, O.~Y., Lee,
506: H.~M., \& Ostriker, J.~P.\ 1999, \apj, 522, 935
507:
508: \bibitem[Gnedin \& Ostriker(1997)]{gnedin_ostriker} Gnedin, O.~Y., \&
509: Ostriker, J.~P.\ 1997, \apj, 474, 223
510:
511: %1996
512: \bibitem[Harris(1996)]{harris} Harris, W.~E.\ 1996, \aj, 112,
513: 1487
514:
515: \bibitem[Irwin et al.(2007)]{irwin} Irwin, M.~J., et al.\
516: 2007, \apjl, 656, L13
517:
518: %1987
519: \bibitem[Inman \& Carney(1987)]{am4} Inman, R.~T., \&
520: Carney, B.~W.\ 1987, \aj, 93, 1166
521:
522: \bibitem[Innanen et al.(1983)]{innanen} Innanen, K.~A., Harris,
523: W.~E., \& Webbink, R.~F.\ 1983, \aj, 88, 338
524:
525:
526: \bibitem[Koposov \& Bartunov(2006)]{sai_cas} Koposov, S.~E., \&
527: Bartunov, O.~S.\ 2006, The Virtual Observatory in Action: New Science, New
528: Technology, and Next Generation Facilities, 26th meeting of the IAU,
529: SPS3, \#19, 3,
530:
531: \bibitem[Koposov et al.(2007)]{koposov} Koposov, S., et al.\
532: 2007, ArXiv e-prints, 706, arXiv:0706.2687
533:
534:
535: % 1998
536: \bibitem[Lindenberg(1998)]{lindenberg} Lindenberg, T., 1998, International Journal of Computer Vision, 30, 2, 79
537:
538: \bibitem[Martin et al.(2007)]{martin} Martin, N.~F., Ibata,
539: R.~A., Chapman, S.~C., Irwin, M., \& Lewis, G.~F.\ 2007, \mnras, 380, 281
540:
541:
542:
543: \bibitem[Meylan \& Heggie(1997)]{meylan97} Meylan, G., \&
544: Heggie, D.~C.\ 1997, \aapr, 8, 1
545:
546: %2003
547: \bibitem[Monet et al.(2003)]{usno} Monet, D.~G., et al.\
548: 2003, \aj, 125, 984
549:
550: \bibitem[Odenkirchen et al.(2001)]{odenkirchen} Odenkirchen, M., et
551: al.\ 2001, \apjl, 548, L165
552:
553:
554: \bibitem[Piotto et al.(2004)]{piotto} Piotto, G., et al.\
555: 2004, \apjl, 604, L109
556:
557: \bibitem[Sarajedini et al.(2006)]{sarajedini} Sarajedini, A., et
558: al.\ 2006, ApJ, submitted, astro-ph/0612598
559:
560: %2002
561: \bibitem[Smith et al.(2002)]{smith} Smith, J.~A., et al.\
562: 2002, \aj, 123, 2121
563:
564: \bibitem[Sosin \& King(1997)]{sosin97} Sosin, C., \& King,
565: I.~R.\ 1997, \aj, 113, 1328
566:
567: \bibitem[Spitzer(1987)]{spitzer} Spitzer, L.\ 1987, Princeton,
568: NJ, Princeton University Press, 1987
569:
570: \bibitem[Stetson(1987)]{daophot} Stetson, P.~B.\ 1987, \pasp,
571: 99, 191
572:
573: \bibitem[Whiting et al.(2002)]{whiting} Whiting, A.~B., Hau,
574: G.~K.~T., \& Irwin, M.\ 2002, \apjs, 141, 123
575:
576: %2005
577: \bibitem[Willman et al.(2005)]{willman} Willman, B., et al.\
578: 2005, \apjl, 626, L85
579:
580: \bibitem[York et al.(2000)]{york} York, D.~G., et al.\ 2000,
581: \aj, 120, 1579
582:
583: \bibitem[Zucker et al.(2006)]{canven} Zucker, D.~B., et al.\
584: 2006, \apjl, 643, L103
585:
586:
587: \end{thebibliography}
588:
589: % I should move the figures back after the bibliography and uncomment the
590: % clearpages...
591:
592:
593:
594:
595:
596:
597: \end{document}
598:
599: