1: %% The first piece of markup in an AASTeX v5.x document
2: %% is the \documentclass command. LaTeX will ignore
3: %% any data that comes before this command.
4:
5: %% The command below calls the preprint style
6: %% which will produce a one-column, single-spaced document.
7: %% Examples of commands for other substyles follow. Use
8: %% whichever is most appropriate for your purposes.
9:
10: \documentclass[12pt,preprint]{aastex}
11:
12: %% manuscript produces a one-column, double-spaced document:
13:
14: %\documentclass[manuscript]{aastex}
15:
16: %% preprint2 produces a double-column, single-spaced document:
17:
18: %%\documentclass[preprint2]{aastex}
19:
20: %% If you want to create your own macros, you can do so
21: %% using \newcommand. Your macros should appear before
22: %% the \begin{document} command.
23: %%
24: %% If you are submitting to a journal that translates manuscripts
25: %% into SGML, you need to follow certain guidelines when preparing
26: %% your macros. See the AASTeX v5.x Author Guide
27: %% for information.
28:
29: \def\steve#1{{\bf[#1---Steve]}}
30: \def\jun#1{{\bf[#1---Jun]}}
31: %\def\steve#1{{}}
32: %\def\jun#1{{}}
33:
34: %% You can insert a short comment on the title page using the command below.
35:
36: %\slugcomment{Jun's version Oct 1, 2002}
37:
38: %% If you wish, you may supply running head information, although
39: %% this information may be modified by the editorial offices.
40: %% The left head contains a list of authors,
41: %% usually a maximum of three (otherwise use et al.). The right
42: %% head is a modified title of up to roughly 44 characters. Running heads
43: %% will not print in the manuscript style.
44:
45: \shorttitle{Central Structure of M15}
46: \shortauthors{Baumgardt et al.}
47:
48: %% This is the end of the preamble. Indicate the beginning of the
49: %% paper itself with \begin{document}.
50:
51: \begin{document}
52:
53: %% LaTeX will automatically break titles if they run longer than
54: %% one line. However, you may use \\ to force a line break if
55: %% you desire.
56:
57: \title{On the central structure of M15}
58:
59: %% Use \author, \affil, and the \and command to format
60: %% author and affiliation information.
61: %% Note that \email has replaced the old \authoremail command
62: %% from AASTeX v4.0. You can use \email to mark an email address
63: %% anywhere in the paper, not just in the front matter.
64: %% As in the title, you can use \\ to force line breaks.
65:
66: \author{Holger Baumgardt\altaffilmark{1},
67: Piet Hut\altaffilmark{2},
68: Junichiro Makino\altaffilmark{1},
69: Steve McMillan\altaffilmark{3},
70: Simon Portegies Zwart\altaffilmark{4}}
71:
72: \altaffiltext{1}{
73: Department of Astronomy, University of Tokyo, 7-3-1 Hongo,
74: Bunkyo-ku,Tokyo 113-0033, Japan}
75:
76: \altaffiltext{2}{
77: Institute for Advanced Study, Princeton, NJ 08540, USA}
78:
79: \altaffiltext{3}{
80: Department of Physics, Drexel University, Philadelphia, PA
81: 19104, USA}
82:
83: \altaffiltext{4}{
84: Astronomical Institute ``Anton Pannekoek,'' University of
85: Amsterdam, Kruislaan 403, 1098 SH Amsterdam, The Netherlands}
86:
87: %% Mark off your abstract in the ``abstract'' environment. In the manuscript
88: %% style, abstract will output a Received/Accepted line after the
89: %% title and affiliation information. No date will appear since the author
90: %% does not have this information. The dates will be filled in by the
91: %% editorial office after submission.
92:
93: \begin{abstract}
94: We present a detailed comparison between the latest observational data
95: on the kinematical structure of the core of M15, obtained with the
96: Hubble STIS and WFPC2 instruments, and the results of dynamical
97: simulations carried out using the special-purpose GRAPE-6 computer.
98: The observations imply the presence of a significant amount of dark
99: matter in the cluster core. In our dynamical simulations, neutron
100: stars and/or massive white dwarfs concentrate to the center through
101: mass segregation, resulting in a sharp increase in $M/L$ toward the
102: center. While consistent with the presence of a central black hole,
103: the Hubble data can also be explained by this central concentration of
104: stellar-mass compact objects. The latter interpretation is more
105: conservative, since such remnants result naturally from stellar
106: evolution, although runaway merging leading to the formation of a
107: black hole may also occur for some range of initial conditions. We
108: conclude that no central massive object is required to explain the
109: observational data, although we cannot conclusively exclude such an
110: object at the level of $\sim500-1000$ solar masses. Our findings are
111: unchanged when we reduce the assumed neutron-star retention fraction
112: in our simulations from 100\% to 0\%.
113:
114: \end{abstract}
115:
116: %% Keywords should appear after the \end{abstract} command. The uncommented
117: %% example has been keyed in ApJ style. See the instructions to authors
118: %% for the journal to which you are submitting your paper to determine
119: %% what keyword punctuation is appropriate.
120:
121: \keywords{black hole physics---globular clusters: individual
122: (M15)---methods: N-body simulations---stellar dynamics}
123:
124:
125: %% From the front matter, we move on to the body of the paper.
126: %% In the first two sections, notice the use of the natbib \citep
127: %% and \citet commands to identify citations. The citations are
128: %% tied to the reference list via symbolic KEYs. The KEY corresponds
129: %% to the KEY in the \bibitem in the reference list below. We have
130: %% chosen the first three characters of the first author's name plus
131: %% the last two numeral of the year of publication as our KEY for
132: %% each reference.
133:
134: \newcommand{\msun}{M_{\odot}}
135: \def\apgt{\ {\raise-.5ex\hbox{$\buildrel>\over\sim$}}\ }
136: \def\aplt{\ {\raise-.5ex\hbox{$\buildrel<\over\sim$}}\ }
137:
138: \section{Introduction}
139:
140: \citet{Gerssenetal2002} have recently reported evidence for an
141: intermediate-mass ($1.7\pm2.7\times10^3\msun$) black hole (IMBH) at
142: the center of globular cluster M15. If confirmed, this would be an
143: exciting and important discovery, and may necessitate a fundamental
144: change in our understanding of the dynamical evolution of globular
145: clusters. To evaluate the need for such a change, we confront the
146: observations with the most detailed cluster simulations currently
147: available.
148:
149: In the standard view \citep{Spitzer1987,MeylanHeggie1997}, globular
150: clusters are born with relatively low central densities. Through
151: two-body relaxation, some of them may reach core collapse, with very
152: high central stellar density. If the cluster contains a significant
153: population ($\apgt10$\%) of primordial binaries, the kinetic energy
154: released by binary--binary and binary--single-star interactions
155: eventually halts the contraction of the core and the cluster reaches a
156: quasi-steady state \citep{GoodmanHut1989} that may endure for
157: substantially longer than a Hubble time. If the cluster contains
158: few primordial binaries, the contraction of the core is halted instead
159: at much higher density by the formation of binaries through
160: three-body interactions. In this case, there is no steady state, and
161: the core may exhibit gravothermal oscillations
162: \citep{Bettwieser1984,Makino1997}.
163:
164: In this picture, the central density of a globular cluster becomes
165: high only after several gigayears, since it typically takes several
166: half-mass relaxation times for core collapse to occur. This view is
167: observationally well supported, since most Galactic globular clusters
168: do have sizeable cores \citep{DjorgovskiMeylan1994,Harris1996}. It is
169: unlikely that an IMBH could have formed as a result of M15's core
170: collapse, as present conditions at the cluster center are unsuitable
171: for runaway stellar collisions to occur
172: \citep{Lee1987,PortegiesZwartMcMillan2002}. Alternative possibilities
173: are that the cluster was initially very compact and that a runaway
174: merger leading to an IMBH may have occurred
175: \citep{PortegiesZwartetal1999}, or that an initial seed black hole
176: grew slowly over a Hubble time via occasional collisions with other
177: stars \citep{MillerHamilton2002}, forming an IMBH by the present time.
178:
179: In this paper we compare the M15 observations with direct $N$-body
180: simulations of star clusters in which stellar evolution and the
181: effects of the Galactic tidal field are realistically taken into
182: account \citep{BaumgardtMakino2002}. In \S2 we describe our cluster
183: model and in \S3, we present ``observations'' of our model cluster and
184: compare them with the actual observations of M15. We briefly
185: summarize and conclude in \S4.
186:
187: \section{Model description}
188:
189: \citet{BaumgardtMakino2002} have performed simulations of star
190: clusters with up to 131072 (128k) stars, using the NBODY4 code
191: \citep{Aarseth1999a} on the GRAPE-6 computer \citep{Makinoetal2002}.
192: Here we concentrate on a member of their ``Family 2.'' Initial
193: stellar masses were chosen from a \citet{Kroupa2001} mass function
194: with lower and upper mass limits of 0.1 and 15 $M_{\odot}$.
195: Primordial binaries were not included. The initial distribution of
196: stars was given by a King model with dimensionless central potential
197: $W_0 = 7$. The model cluster was placed on a circular orbit at a
198: distance of 8.5 kpc from the Galactic center. The Galactic potential
199: was treated as a singular isothermal sphere with a constant rotation
200: velocity of 220 km/s. Stellar evolution was modeled according to
201: \citet{Hurleyetal2000}. The initial half-mass radius of the cluster
202: (with $N=128$k stars and a mass of $7.2\times 10^4$ M$_\odot$) was 7.1
203: pc; the initial half-mass crossing time was 4.1 Myr. Core collapse
204: occurred at $T=12.6$ Gyr, when the remaining cluster mass was
205: $\sim2\times 10^4$ M$_\odot$. The calculation, to the point of
206: complete dissolution, took about 1000 hours computing time on a
207: 4-board, single-host GRAPE-6 system. Details of the calculation are
208: described in \citet{BaumgardtMakino2002}.
209:
210: Note that our 128k-body model still contains far fewer stars than
211: M15---we cannot yet perform star-by-star simulations of a relatively
212: large globular cluster. Rather, we compare nondimensional quantities,
213: such as the radial dependence of the velocity dispersion, its slope,
214: $M/L$ etc. In the next section we present a comparison of the
215: luminosity and velocity dispersion profiles near the centers of the
216: two systems.
217:
218: In the calculations of \citet{BaumgardtMakino2002}, collisions between
219: stars were not taken into account, and hence massive black holes could
220: not form. We have also performed simulations in which stellar
221: collisions were properly included
222: \citep{PortegiesZwartetal2001,PortegiesZwartMcMillan2002} and find
223: that, for initial conditions appropriate for globular clusters, the
224: neglect of stellar collisions is justified.
225:
226: \section{Analysis}
227:
228: Figure \ref{fig1} shows the ``observed'' line-of-sight velocity
229: dispersion profile of our model cluster. In order to improve
230: statistics, we have superimposed ten snapshots spanning a 500 Myr
231: period following core collapse. We calculated the velocity dispersion
232: of the model cluster in two ways. First, we determined the velocity
233: dispersion using all stars (including compact remnants), averaging
234: over three orthogonal directions. In the second method, we used only
235: stars brighter than $V=19$ at the distance of M15 (assumed to be 10
236: kpc), the sample actually used by \citet{Gerssenetal2002}. Except for
237: the innermost parts, both profiles agree rather well with each other;
238: within the error bars, the model velocity dispersion profile is also
239: very similar to that of M15 (Figure 9 of Gerssen et al. 2002).
240: \begin{figure}[htbp!]
241: \plotone{f1.eps}
242: \caption{Line-of-sight velocity dispersion, $\sigma_{\rm LOS}$, as a
243: function of projected distance from the cluster center. Ten snapshots
244: with time intervals of 50 Myr are overlaid to improve statistics.
245: Velocity dispersions are averaged over orientation angles. Crosses
246: are calculated using all stars in the cluster, and filled circles
247: using stars with visual magnitude $V < 19$ at 10 kpc. The upper axis
248: gives distances in arcseconds, calculated by assuming that our model
249: cluster is observed from a distance of 10 kpc.
250: \label{fig1}}
251: \end{figure}
252:
253: % Theoretically, this is a quite natural behavior for a cluster in deep
254: % collapse. In case of an idealized cluster made of equal-mass
255: % particles, velocity dispersion would be expressed as $\sigma \propto
256: % r^{-0.13}$, if we take the density profile as $\rho \propto r^{-2.26}$
257: % \citep{Baumgardtetal2002}. This gives a 50\% increase of velocity
258: % dispersion for a change in radius by a factor of 30. With realistic
259: % multi-mass clusters, the slope of the velocity would become even
260: % smaller due to mass segregation.
261:
262: Figure \ref{fig2} depicts the surface number density of bright
263: ($V<22$) stars and of compact remnants. The adopted cutoff of $V=22$
264: is the photometric limit found in the study of the cluster center by
265: \citet{SosinKing1997}, and is also consistent with the limit of
266: $V=22.5$ in the data of \citet{vandermareletal2002}. For both groups,
267: the inner region shows clear power-law cusps, with indices of
268: approximately $-0.8$ and $-1.2$, respectively. The surface density of
269: bright stars is again in very good agreement with the HST WFPC2 and
270: FOC star count results \citep{Guhathakurtaetal1996,SosinKing1997}.
271:
272: \begin{figure}[htbp!]
273: \plotone{f2.eps}
274: \caption{Radial surface number density profiles for different stellar
275: groups just after core collapse. Open circles denote white dwarfs and
276: neutron stars. Filled triangles denote stars with $V<22$. In the
277: center, the slope for bright stars is similar to that observed in M15.
278: The bright stars follow a much shallower distribution than the compact
279: remnants, due to mass segregation.
280: \label{fig2}}
281: \end{figure}
282:
283: A cluster in deep collapse should have a density profile steeper than
284: isothermal, since the velocity dispersion increases inward
285: \citep{LBE1980, Cohn1980}. One might wonder why the central density
286: profile of bright stars shows a slope shallower than that of an
287: isothermal sphere ($\sim r^{-1}$ in projection). The reason is simply
288: that the bright stars are not the most massive components in
289: present-day globular clusters \citep{MurphyCohn1988, Luggeretal1995}.
290: Compact remnants (neutron stars and massive white dwarfs) are more
291: massive, and are the dominant population in the central region (see
292: Figure\,\ref{fig2}). Their density profile ($\rho \sim r^{-2.2}$ in 3
293: dimensions) is close to the theoretical prediction for the central
294: profile of a core-collapsed cluster: $\rho \sim r^{-2.26}$ (see
295: \citet{Baumgardtetal2002} and references therein). As a consequence,
296: proper interpretation of Figure\,\ref{fig1} must take into account the
297: substantial radial variation of the mass to light ratio in the cluster
298: core.
299:
300: \begin{figure}[htbp!]
301: \plotone{f3.eps}
302: \caption{Line-of-sight velocity dispersion of the $V<19$ stars in the
303: $N$-body simulations (filled circles), and inferred from the stellar
304: number density and cluster potential (solid and dashed curves). The
305: solid curve shows the inferred velocity dispersion of stars with
306: $V<22$, using the potential calculated from all stars. Dashed curves
307: are calculated using the potential determined from stars with $V<22$,
308: assuming a constant $M/L$, together with central point masses of
309: (bottom to top) 0, 40, 80 and 120 $M_\sun$. The value of $M/L$ is
310: chosen to fit the measured velocity dispersion between 1 and 10 pc
311: from the cluster center. For constant assumed $M/L$, the best fit has
312: $M_{\rm BH} \sim 80M_\sun$.
313: \label{fig3}}
314: \end{figure}
315:
316: As an illustration, we compare the observed velocity dispersion
317: profile of our model with the velocity dispersion profile inferred
318: from the distribution of bright stars, using the Jeans equation with
319: an isotropic velocity distribution and a constant mass-to-light ratio.
320: The numerical procedure is as described by \citet{Gerssenetal2002}
321: (section 5). Figure \ref{fig3} shows the result. Not surprisingly,
322: we find a large discrepancy between the inferred velocity dispersion
323: and the observed profile, as illustrated by the lowest dashed line in
324: Figure 3. The predicted central velocity dispersion, based on the
325: mass contribution of the visible stars, would actually dip in the
326: center, contrary to what is observed. Most of the discrepancy is
327: caused by the neglect of the central concentration of dark matter in
328: the form of stellar remnants. Trying to improve the fit by
329: introducing a central point mass as a free parameter leads to a
330: central mass of approximately 80 $\mbox{M}_{\odot}$ (second dashed
331: line from the top in Figure\,\ref{fig3}).
332:
333: \citet{Gerssenetal2002} have analysed the velocity distribution of the
334: bright stars in M15, using two different methods and averaging the
335: results. They first assume a constant mass-to-light ratio, then adopt
336: a more realistic radial run of mass to light, obtained from
337: Fokker-Planck simulations \citep{Dulletal1997}. Their first method
338: leads to an inferred central black hole mass of $3.2 \times 10^3 \;
339: \mbox{M}_\odot$, containing a fraction of $3.2 \times 10^3 \;
340: \mbox{M}_\odot/4.9 \times 10^5 \; \mbox{M}_\odot = 0.65\%$ of the
341: total cluster mass. (The choice of M15 mass is taken from Dull {\em
342: et al.}) This is similar to the fractional mass of the central point
343: mass deduced above from Figure \ref{fig3}, to which we ascribed a mass
344: ratio of $80 \; \mbox{M}_\odot/20,000 \; \mbox{M}_\odot = 0.4\%$ of
345: our cluster mass.
346:
347: Using the correct cluster potential (solid line in Figure\,\ref{fig3})
348: in the analysis of our simulations recovers the velocity dispersion of
349: stars of $V<19$ without the need for a central point mass. In effect,
350: we use the (known) variation in the mass-to-light ratio of the model
351: cluster to convert from the observed $V<22$ number density to the
352: actual potential. Comparison of the central point-mass data with the
353: error bars in the ``observed'' ($V<19$) velocity dispersion in
354: Figure\,\ref{fig3} suggests that the largest point mass that could be
355: hidden in the data has a mass of $\aplt 40 M_\odot$. This would
356: correspond to $\aplt 10^3 M_\odot$ in the M15 system.
357: %
358: In contrast, the equivalent (second) method employed by
359: \citet{Gerssenetal2002} actually increased the inferred central mass
360: to $4.5 \times 10^3 \; \mbox{M}_\odot$. This mass determination was
361: subsequently shown to be erroneous \citep{Gerssenetal2003}.
362: %
363: \footnote{\citet{Gerssenetal2003} report that Figures 9 and 12 of Dull
364: {\em al.} contained errors which critically affected their analysis.
365: They were already aware of this fact, and informed us of it after
366: receiving a copy of the submitted manuscript of the current paper.}
367: %
368: The correct treatment yielded a formal mass of $1.7 \times 10^3 \;
369: \mbox{M}_\odot$; however a mass of zero was excluded only at the
370: $\sim\frac12\,\sigma$ level.
371:
372: In their addendum, \citep{Gerssenetal2003} maintain that a central
373: black hole remains a viable interpretation of the M15 data, citing the
374: probability that most neutron stars would have escaped the cluster on
375: formation, in contradiction to the assumption of 100\% neutron star
376: retention made by \citet{Dulletal1997} and also in Figure 3 above.
377: Most neutron stars receive substantial ``kicks'' at birth
378: \citep{LyneLorimer1994}, which may eject them from their parent
379: cluster. Theoretical estimates of the retention fraction range from
380: $\sim5$ to $\sim$20\% \citep{Drukier1996}. If no neutron stars were
381: present in the core, the slope of the luminosity profile would be
382: expected to steepen somewhat \citep{TakahashiLee2000}.
383:
384: To address this possibility, we have repeated our earlier simulation
385: with the extreme alternative assumption that no neutron stars were
386: retained. The result is plotted in Figure 4, which presents the
387: analogous information to Figure 3 for this model. The discrepancy
388: between the ``observed'' velocity profile and the expected profile
389: calculated from the distribution of stars with $V<22$, assuming a
390: constant mass-to-light ratio, still exists, since in this case massive
391: white dwarfs have accumulated in the center and replaced the
392: main-sequence stars. Thus, changing the neutron star retention
393: fraction does not significantly alter our conclusion. The assumption
394: of constant $M/L$ now yields a black-hole mass of $\sim40M_\sun$, half
395: the value found in Figure 3, since massive white dwarfs and neutron
396: stars contributed roughly equally to the central dark mass in that
397: model.
398:
399: Using the same reasoning as before, we then estimate a value of
400: $20M_\sun$, for the maximum mass that could be hidden in the form of a
401: central black hole. In the case of M15, this would correspond to
402: $\aplt 500 M_\odot$. We note, however, that the fitting procedure is
403: relatively insensitive to the precise nature of the dark matter
404: contained within the innermost 0.5 pc (H. Cohn and P. Lugger, 2002,
405: private communication). The present data are probably consistent with
406: dark matter in the form of a range of combinations of neutron stars,
407: massive white dwarfs, or an intermediate-mass black hole.
408:
409: \begin{figure}[htbp!]
410: \plotone{f4.eps}
411: \caption{As for Figure 3, but for a model with 0\% neutron star
412: retention. For constant $M/L$, the best-fitting black-hole mass is
413: now $40M_\sun$.
414: \label{fig4}}
415: \end{figure}
416:
417: \section{Conclusions}
418:
419: In this paper we compare recent observations of the central regions of
420: M15 with recent direct $N$-body simulations of realistic models of
421: star clusters. We find that the velocity dispersion and luminosity
422: profiles obtained from the $N$-body simulations, after appropriate
423: scaling, reproduce the observations without any need to invoke a
424: central point mass. Earlier Fokker--Planck results without a black
425: hole \citep{Dulletal1997, Dulletal2002} are also consistent with the
426: current observations. Thus we conclude that the M15 observations can
427: be adequately explained without recourse to a central massive black
428: hole.
429:
430: Although the current observations do not prove the existence of a
431: central massive black hole, they do not disprove it either. Our
432: analysis (Figure\,\ref{fig3}) indicates that a moderate
433: intermediate-mass black hole of $\sim10^3$ solar masses is still
434: possible. Such an object is not altogether unexpected, since it might
435: have formed early in the cluster's evolution through runaway merging
436: \citep{Ebisuzakietal2001,PortegiesZwartMcMillan2002}. To confirm this
437: interesting possibility, or to place more stringent limits on the mass
438: of a possible black hole, will require detailed evolutionary modeling
439: of the cluster for different evolutionary scenarios. We plan to carry
440: out such simulations in the near future.
441:
442: \section*{Acknowledgments}
443:
444: HB and JM thank Toshi Fukushige and Yoko Funato for stimulating
445: discussions; PH and SM acknowledge several collegial conversations
446: with Roeland van der Marel, Karl Gebhardt, Haldan Cohn and Phyllis
447: Lugger following submission of this paper. We also acknowledge
448: detailed and helpful comments on the manuscript by Phyllis Lugger and
449: an anonymous second referee. This work is supported in part by
450: Grant-in-Aid for Scientific Research B (13440058) of the Ministry of
451: Education, Culture, Sports, Science and Technology, Japan, by grants
452: NASA ATP grants NAG5-6964 and NAG5-9264, and by the Royal Netherlands
453: Academy of Sciences (KNAW) and the Netherlands Research School for
454: Astronomy (NOVA)
455:
456: \begin{thebibliography}{}
457:
458: \bibitem[Aarseth(1999)]{Aarseth1999a}
459: Aarseth, S. J. 1999, \pasp, 111, 1333
460: \bibitem[Baumgardt et al.(2002)]{Baumgardtetal2002}
461: Baumgardt, H., Heggie, D., Hut, P., and Makino, J. 2002, submitted to \mnras
462: \bibitem[Baumgardt and Makino(2002)]{BaumgardtMakino2002}
463: Baumgardt, H., and Makino, J. 2002, to appear in {\mnras} ({\tt
464: astro-ph/xxxxx})
465: \bibitem[Bettwieser and Sugimoto(1984)]{Bettwieser1984}
466: Bettwieser, E., and Sugimoto, D. 1984, \mnras, 208, 493
467: \bibitem[Cohn(1980)]{Cohn1980}
468: Cohn, H. 1980, \apj, 242, 765
469: \bibitem[Djorgovski and Meylan(1994)]{DjorgovskiMeylan1994}
470: Djorgovski, S., and Meylan, G. 1994, \aj, 108, 1292
471: \bibitem[Drukier(1996)]{Drukier1996}
472: Drukier, G. A. 1996, \mnras, 280, 498
473: \bibitem[Dull et al.(1997)]{Dulletal1997}
474: Dull, J. D., Cohn, H. N., Lugger, P. M., Murphy, B. W., Seitzer, P. O.,
475: Callanan, P. J., Rutten, R. G. M., and Charles, P. A. 1997, \apj, 481, 267
476: \bibitem[Dull et al.(2002)]{Dulletal2002}
477: Dull, J. D., Cohn, H. N., Lugger, P. M., Murphy, B. W., Seitzer, P. O.,
478: Callanan, P. J., Rutten, R. G. M., and Charles, P. A. 2002, submitted
479: to {\apj} ({\tt astro-ph/0210588})
480: \bibitem[Ebisuzaki et al.(2001)]{Ebisuzakietal2001}
481: Ebisuzaki, T., Makino, J., Tsuru, T. G., Funato, Y., Portegies Zwart, S. F.,
482: Hut, P., McMillan, S. L. W., Matsushita, S., Matsumoto, H., and Kawabe, R.
483: 2001, \apjl, 562, L19
484: \bibitem[Gerssen et al.(2002)]{Gerssenetal2002}
485: Gerssen, J., van der Marel, R. P., Gebhardt, K., Guhathakurta, P.,
486: Peterson, R., and Pryor, C. 2002, \aj, in press
487: \bibitem[Gerssen et al.(2003)]{Gerssenetal2003}
488: Gerssen, J., van der Marel, R. P., Gebhardt, K., Guhathakurta, P.,
489: Peterson, R., and Pryor, C. 2003, \aj, in press
490: \bibitem[Goodman and Hut(1989)]{GoodmanHut1989}
491: Goodman, J., and Hut, P. 1989, \nat, 339, 40
492: \bibitem[Guhathakurta et al.(1996)]{Guhathakurtaetal1996}
493: Guhathakurta, P., Yanny, B., Schneider, D. P., and Bahcall, J. N. 1996,
494: \aj, 111, 267
495: \bibitem[Harris(1996)]{Harris1996}
496: Harris, W. E. 1996, \aj, 112, 1487 ({\tt
497: http://physun.physics.mcmaster.ca/Globular.html})
498: \bibitem[Hurley et al.(2000)]{Hurleyetal2000}
499: Hurley, J. R., Pols, O. R., and Tout, C. A. 2000, \mnras, 315, 543
500: \bibitem[Kroupa(2001)]{Kroupa2001}
501: Kroupa, P. 2001, \mnras, 322, 231
502: \bibitem[Lee(1987)]{Lee1987}
503: Lee, H. M. 1987, \apj, 319, 801
504: \bibitem[Lugger et al.(1995) ]{Luggeretal1995}
505: Lugger, P. M., Cohn, H. N., \& Grindlay, J. E. 1995, \apj, 439, 191
506: \bibitem[Lynden-Bell \& Eggleton(1980)]{LBE1980}
507: Lynden-Bell, D., \& Eggleton, P. P. 1980, \mnras, 191, 483
508: \bibitem[Lyne \& Lorimer(1994)]{LyneLorimer1994}
509: Lyne, A. G., Lorimer, D. R. 1994, Nature, 369, 127
510: \bibitem[Makino(1997)]{Makino1997}
511: Makino, J. 1997, \apj, 478, 58
512: \bibitem[Makino et al.(2002)]{Makinoetal2002}
513: Makino, J., Fukushige, T., and Namura, K. 2002, in preparation
514: \bibitem[van der Marel et al.(2002)]{vandermareletal2002}
515: van der Marel, R. P., Gerssen, J., Guhathakurta, P., Peterson, R. C., and
516: Gebhardt, K. 2002, \apj, in press
517: \bibitem[Meylan and Heggie(1997)]{MeylanHeggie1997}
518: Meylan, G., and Heggie, D. C. 1997, \aapr, 8, 1
519: \bibitem[Miller and Hamilton(2002)]{MillerHamilton2002}
520: Miller, M. C., and Hamilton, D. P. 2002, \mnras, 330, 232
521: \bibitem[Murphy \& Cohn(1988)]{MurphyCohn1988}
522: Murphy, B. W., \& Cohn, H. N. 1988, \mnras, 232, 835
523: \bibitem[Portegies Zwart et al.(1999)]{PortegiesZwartetal1999}
524: Portegies Zwart, S. F., Makino, J., McMillan, S. L. W., and Hut,
525: P. 1999, \aap, 348, 117
526: \bibitem[Portegies Zwart et al.(2001)]{PortegiesZwartetal2001}
527: Portegies Zwart, S. F., McMillan, S. L. W., Hut, P., and Makino,
528: J. 2001, \apj, 546, L101
529: \bibitem[Portegies Zwart and McMillan(2002)]{PortegiesZwartMcMillan2002}
530: Portegies Zwart, S. F., and McMillan, S. L. W. 2002, \apj, 576, 899
531: \bibitem[Sosin and King(1997)]{SosinKing1997}
532: Sosin, C., and King, I. R. 1997, \aj, 113, 1328
533: \bibitem[Spitzer(1987)]{Spitzer1987}
534: Spitzer, L. J. 1987, {\em Dynamical Evolution of Globular Clusters}.
535: Princeton University Press, Princeton, New Jersey.
536: \bibitem[Takahashi \& Lee(2000)]{TakahashiLee2000}
537: Takahashi, K. and Lee, H. M. 2000, \mnras, 316, 671
538:
539: \end{thebibliography}
540:
541: \end{document}
542: