1: \documentclass[12pt,preprint]{aastex}
2: \shorttitle{GLOBULAR CLUSTER BLUE STRAGGLERS}
3: \shortauthors{Piotto et al.}
4:
5: \begin{document}
6:
7: \def\hst{{\sl HST}}
8:
9: \title{Relative Frequencies of Blue Stragglers in Galactic Globular
10: Clusters:\\Constraints for the Formation Mechanisms\footnote{Based on
11: observations with the NASA/ESA {\it Hubble Space Telescope}, obtained at
12: the Space Telescope Science Institute, which is operated by AURA, Inc.,
13: under NASA contract NAS 5-26555.}}
14:
15: \author{Giampaolo Piotto\altaffilmark{1},
16: Francesca De Angeli\altaffilmark{1},
17: Ivan R.\ King\altaffilmark{2},
18: S.\ G.\ Djorgovski\altaffilmark{3},
19: Giuseppe Bono\altaffilmark{4},
20: Santi Cassisi\altaffilmark{5},
21: Georges Meylan\altaffilmark{6},
22: Alejandra Recio-Blanco\altaffilmark{1},
23: R.\ M.\ Rich\altaffilmark{7},
24: Melvyn B. Davies\altaffilmark{8}
25: }
26:
27: \altaffiltext{1}{Dipartimento di Astronomia, Universit\`a
28: di Padova, Vicolo dell'Osservatorio 2, I-35122 Padova, Italy;
29: piotto@pd.astro.it, deangeli@pd.astro.it, recio@pd.astro.it}
30:
31: \altaffiltext{2}{Department of Astronomy, University of Washington,
32: Box 351580, Seattle, WA 98195-1580; king@astro.washington.edu}
33:
34: \altaffiltext{3}{California Institute of Technology, MS 105-24,
35: Pasadena, CA 91125; george@astro.caltech.edu}
36:
37: \altaffiltext{4}{Osservatorio Astronomico di Roma, Via Frascati 33,
38: 00040 Monte Porzio Catone, Italy; bono@mporzio.astro.it}
39:
40: \altaffiltext{5}{Osservatorio Astronomico di Collurania, via M. Maggini,
41: 64100 Teramo; cassisi@astrte.te.astro.it}
42:
43: \altaffiltext{6}{Space Telescope Science Institute, 3700 San Martin Drive,
44: Baltimore MD 21218, U.S.A., gmeylan@stsci.edu}
45:
46: \altaffiltext{7}{Department of Physics and Astronomy, Division of
47: Astronomy
48: and Astrophysics, University of California, Los Angeles, CA 90095-1562;
49: rmr@astro.ucla.edu}
50:
51: \altaffiltext{8}{Department of Physics and Astronomy, University of
52: Leicester, Leicester, LE1 7RH, UK}
53:
54: \begin{abstract}
55: We discuss the main properties of the Galactic globular cluster (GC)
56: blue straggler stars (BSS), as inferred from our new catalog containing
57: nearly 3000 BSS. The catalog has been extracted from the photometrically
58: homogeneous $V$ vs. ($B-V$) color-magnitude diagrams (CMD) of 56 GCs,
59: based on WFPC2 images of their central cores. In our analysis we used
60: consistent relative distances based on the same photometry and
61: calibration. The number of BSS has been normalized to obtain relative
62: frequencies ($F_{\rm BSS}$) and specific densities ($N_{\rm S}$) using
63: different stellar populations extracted from the CMD. The cluster
64: $F_{\rm BSS}$ is significantly smaller than the relative frequency of
65: field BSS. We find a significant anti-correlation between the BSS
66: relative frequency in a cluster and its total absolute luminosity
67: (mass). There is no statistically significant trend between the BSS
68: frequency and the expected collision rate.
69: $F_{\rm BSS}$ does not depend on other cluster parameters, apart from a
70: mild dependence on the central density. PCC clusters act like normal
71: clusters as far as the BSS frequency is concerned. We also show that
72: the BSS luminosity function for the most luminous clusters is
73: significantly different, with a brighter peak and extending to brighter
74: luminosities than in the less luminous clusters. These results imply
75: that the efficiency of BSS production mechanisms and their relative
76: importance vary with the cluster mass.
77: \end{abstract}
78:
79: \keywords{stars: blue stragglers --- globular clusters: general ---
80: stars:luminosity function --- Hertzsprung-Russell diagram}
81:
82: \section{Introduction}
83: Globular Clusters (GCs)
84: are important astrophysical laboratories for
85: investigating the stellar dynamics and stellar evolution of low-mass
86: stars
87: (e.g., Meylan \& Heggie 1997). In recent years, it became
88: clear that we can not study these two astrophysical processes
89: independently if we want to understand GCs and properly address
90: several long-standing problems concerning their stellar content.
91:
92: Among the most puzzling products of the interplay between stellar
93: evolution and dynamics are the blue straggler stars (BSS). This group of
94: stars was originally identified by Sandage (1953) in the cluster M3 as a
95: bluer and brighter extension of the main sequence (MS) turn-off (TO)
96: stars. At present, the most popular mechanisms suggested to account for
97: their origin are {\em primordial binary evolution} (McCrea 1964), i.e.,
98: mass transfer and/or coalescence in primordial binary systems (Carney et
99: al.\ 2001), and {\em collisional merging}, i.e., the collision of single
100: and/or binary systems (Bailyn 1995). Unfortunately, current photometric
101: investigations do not allow us to figure out the mechanism that triggers
102: the formation of BSS in GCs, and indeed it has been suggested that both
103: primordial binary evolution and collisions are probably at work in
104: different clusters (Ferraro, Fusi Pecci, \& Bellazzini 1995; Piotto et
105: al.\ 1999, Ferraro et al. 2003), or even within the same cluster
106: (Ferraro et al.\ 1997). The observational scenario concerning BSS
107: formation has been recently enriched by the results of a spectroscopic
108: survey by Preston \& Sneden (2000, hereafter PS00). On the basis of
109: multi-epoch radial velocity data of field blue metal-poor (BMP) stars,
110: PS00 found that more than 60\% of the stars in their sample are
111: binaries. On the basis of empirical evidence, PS00 concluded that at
112: least 50\% of BMP stars are BSS. Moreover, PS00 suggested that the BSS
113: in their sample must have formed via mass transfer in binaries.
114: Finally, PS00 found that the specific frequency of BSS in the local halo
115: is an order of magnitude larger than in GCs. This discrepancy
116: opens several new questions concerning the origin of
117: field and cluster BSS.
118:
119: In an attempt to better understand the properties of BSS stars in GCs,
120: we took advantage of our homogeneous database of color-magnitude
121: diagrams (CMD) from WFPC2 images (Piotto et al.\ 2002) to select
122: a sample of nearly 3000
123: BSS in 56 GCs characterized by different morphological and
124: dynamical properties.
125: In
126: this paper we exploit the new BSS catalog to investigate
127: empirically whether the BSS population is related to any of the
128: properties of the parent GC. Here we present the results we
129: believe to be the most relevant and original. The entire BSS photometric
130: catalog and further details on the BSS extraction
131: will be published in
132: a forthcoming paper (De Angeli et al.\ 2004), and it
133: will become available at the Padova Globular Cluster Group web site.
134:
135: \section{THE DATABASE OF BSS IN GCs}
136: \label{database}
137: We have recently completed our HST/WFPC2 snapshot project (GO 7470, GO
138: 8118, GO 8723). By adding the data from our former GO 6095, and
139: similar data from the archive (i.e., WFPC2 images collected in the
140: cluster center, in the F439W and F555W bands), we have obtained a total of
141: 74 CMDs. They can be found on the Padova Globular Cluster Group web
142: pages (http://dipastro.pd.astro.it/globulars). Details on the data
143: reduction are in Piotto et al.\ (2002). Here suffice it to say that all
144: the data have been processed in the same way: instrumental photometry
145: with DAOPHOT/ALLFRAME, CTE correction and calibration
146: to the $B$ and $V$ standard systems following Dolphin
147: (2000).
148: Artificial star experiments have been run for all the clusters to
149: estimate the completeness of star counts. Completeness corrections have
150: been always applied when necessary. For the BSS sample the completeness
151: was typically larger than 90\%.
152:
153: BSS are present in all 74 GCs of our catalog. However, 18 CMDs were
154: too contaminated by background/foreground stars, or heavily affected by
155: differential reddening, to allow us a reliable selection of BSS.
156: The final catalog,
157: includes 2798 BSS
158: candidates in 56 GCs (actual stars, before completeness correction), i.e., about
159: five times the number published in previous catalogs (Fusi
160: Pecci et al.\ 1993; Sarajedini 1993).
161:
162: The BSS in our sample share one common feature: in the inner region
163: mapped by our WFPC2 images, the radial distribution of BSS is more
164: centrally peaked than that of any other cluster population. In fact, we
165: performed several Kolmogorov-Smirnov tests on the radial
166: distributions of HB, RGB, and BSS, and we found that the BSS are
167: more centrally concentrated at the level of $99.9$\%, apart
168: from a few cases where the small number of stars prevents any
169: statistically significant test.
170:
171: \section{THE RELATIVE FREQUENCY OF BSS}
172: \label{freq}
173:
174: In this section, we investigate whether the number of BSS is correlated
175: with any of the physical and morphological parameters of their parent
176: GCs. In order to compare the number of BSS in different GCs properly, we
177: must adjust it to allow for how much of each cluster we sampled. To
178: this end, a number of different specific frequencies (defined as the
179: ratio between the number of BSS and a reference population) have been
180: used in the literature (cf. Ferraro et al.\ 1995).
181: We estimated the specific frequencies by normalizing the number of
182: BSS to the HB ($F^{\rm HB}_{\rm BSS}$) or the RGB ($F^{\rm RGB}_{\rm
183: BSS}$) stars. Interestingly, the results discussed below do not depend
184: on which specific frequency we choose. Therefore, in the following we
185: will adopt $F^{\rm HB}_{\rm BSS}$, and will refer to it as $F_{\rm
186: BSS}$. The numbers of BSS and HB stars have been corrected for
187: completeness
188: (details in Piotto et al.\ 2002)
189: before calculating
190: $F_{\rm BSS}$.
191:
192: \begin{figure}
193: \plotone{f1.ps}
194: \figcaption{BSS relative frequency as a function of the integrated absolute
195: magnitude of the cluster ({\it top panel}), the central density
196: ({\it middle panel}), and of the
197: collision rate ({\it bottom panel}).
198: Different symbols are used (in the top and bottom panels) for
199: clusters with different central densities:
200: $\log\rho_0<2.8:$ {\it open triangles};
201: $2.8<\log\rho_0<3.6:$ {\it filled triangles};
202: $3.6<\log\rho_0<4.4:$ {\it crosses};
203: $\log\rho_0>4.4:$ {\it filled circles}.
204: In all panels, PCC clusters are open circles.
205: }
206: \end{figure}
207:
208: The top panel of Fig.\ 1 shows $F_{\rm BSS}$ as a function of
209: the integrated visual absolute magnitude of the cluster,
210: from the integrated visual apparent magnitudes and the
211: reddening values in the Harris catalog and the apparent distance
212: moduli derived by Recio-Blanco et al.\ (2004a) for the same
213: clusters by following the procedure outlined by Zoccali et al.\ (2000).
214: Even though the distribution of empirical data still shows a
215: large scatter for $-8 \le M_V \le -7.4$, the data in this panel
216: clearly show a correlation between $F_{\rm BSS}$ and the integrated
217: absolute
218: magnitude $M_V$. In particular, the faintest clusters in our sample,
219: namely NGC 6717 and NGC 6838, present a BSS specific frequency that is
220: more than a factor of 20 larger than that for the brightest clusters.
221: This result is in agreement with that found by PS00 on the basis of data
222: collected from the literature for 30 GCs
223: though their plot tends to
224: flatten for $M_V<-7$, while Fig.\ 1 shows that $F_{\rm BSS}$ decreases
225: continuously with increasing cluster total luminosity, up to $M_V<-9$.
226: Interestingly enough, PCC clusters (open circles), with the
227: exception of NGC 5946 that shows a small value of $F_{\rm BSS}$,
228: behave
229: as normal clusters.
230:
231: The middle panel of Fig.\ 1 shows that $F_{\rm BSS}$ also depends on
232: the cluster central density $\rho_0$ (in $L_\odot/pc^3$), though this
233: correlation is statistically somewhat less significant than the
234: previous one. (For internal consistency, we have re-calculated the
235: central densities using the equations suggested by Djorgovski (1993),
236: but adopting our new distance moduli and the central surface
237: brightness in the Harris catalog.) The $F_{\rm BSS}$ for clusters
238: with $\log\rho_0>3.2$ shows a large dispersion, and no correlation.
239: For $\log\rho_0<3.2$, the BSS frequency
240: increases with decreasing central density.
241: Once again, the PCC clusters do not
242: show any peculiar trend.
243: We have
244: also compared $F_{\rm BSS}$ with the concentration parameter $c$ and the
245: half mass relaxation time $t_{\rm h}$. Here too there is no clear correlation,
246: though GCs with $\log t_{\rm h}<9$ have, on average, a $F_{\rm BSS}$
247: three times larger than clusters with a longer relaxation time.
248:
249: In view of the proposed formation mechanisms for BSS, it is interesting
250: to check whether
251: the $F_{\rm BSS}$
252: depends on the expected
253: frequency of stellar collisions. King (2002) demonstrated that the rate
254: of stellar collisions (per cluster and per year) in a King model GC is
255: about $\Gamma_{\rm c} = 5\times10^{-15}(\Sigma_0^3r_{\rm c})^{1/2}$,
256: where $\Sigma_0$ is the central surface brightness in units of $L_{\odot
257: V}$ pc$^{-2}$ (equivalent to $\mu_v=26.41$), and $r_{\rm c}$ is the core
258: radius in parsecs (taken from Harris 1996). In order to calculate the
259: probability $\Gamma_\star$ that a given star will have a collision in
260: one year we have divided this collision rate by the total
261: number of stars ($N_{\rm star}$) in the cluster. This has been estimated
262: by using the integrated visual absolute magnitude of the cluster,
263: assuming $M/L=2$ and a typical mass for the colliding stars of $0.4
264: m_\odot$. The bottom panel of Figure 1 plots $F_{\rm BSS}$ as a
265: function of the resulting $\Gamma_\star$. There is no statistically
266: significant correlation, though we note that, on average, the
267: 11
268: clusters with the smallest collision probability
269: ($\Gamma_\star<10^{-15}$) have a BSS frequency 2--3 times higher than
270: clusters with higher collision rates. It must be noted here that,
271: according to the results by PS00, the BSS frequency
272: of the GCs with the smallest collision probability in our sample is
273: about 5 times smaller than the BSS frequency in the field, where
274: collisions are so much rarer.
275: We note that in the case of PCC clusters, the current values
276: $\Gamma_\star$ and $\rho_0$ may not be representative of the average
277: dynamical environment in which currently--observed blue stragglers
278: have formed. However, Fig\ 1 shows that PCC clusters have BSS
279: frequencies comparable to normal King--model clusters, possibly
280: indicating that their dynamical evolution has not affected in a
281: significant way the BSS formation.
282:
283: The anti-correlation between $F_{\rm BSS}$ and total cluster
284: luminosity, the lack of a statistically significant correlation
285: with the collisional parameter, and the apparently higher relative
286: frequency of BSS where collision rates are very small are the
287: most interesting results extracted from our catalog. These empirical
288: facts are somehow puzzling. In fact, we would have expected more BSS in
289: clusters where the probability of collision is higher. We will discuss
290: these results further in the next Section.
291:
292: The error bars plotted in Fig.\ 1 account for Poisson sampling errors
293: and the uncertainty in the completeness corrections. Even if we assume
294: an upper limit of 0.2 magnitude for the uncertainties in the individual
295: distance moduli, the correlation shown in Fig.\ 1 represents a robust
296: empirical result, and the reasons are manifold. Unlike other data
297: available in the literature, i) our dataset is photometrically
298: homogeneous; ii) the star counts included the 2 innermost
299: arcmin; iii) the star counts of BSS and of the reference populations
300: have been corrected for incompleteness. As a whole, the present dataset
301: is only very marginally affected by the thorny statistical problems
302: affecting previous estimates of BSS specific frequencies (cf.\
303: discussion in Ferraro et al.\ 1995).
304:
305: \section{DISCUSSION}
306: \label{conc}
307:
308: The statistically significant anticorrelation of the BSS relative
309: frequency with the integrated luminosity and the
310: independence of
311: the expected collision rate discussed in the previous Section are
312: noteworthy, and we will concentrate on them. These observational facts
313: are complemented by the finding by PS00 that field
314: BSS have a frequency $F_{\rm BSS}=4.0$,
315: an order of magnitude larger than the BSS
316: frequency of the bulk of the GCs.
317:
318: \begin{figure}
319: \plotone{f2.ps}
320: \figcaption{Number of BSS (circles), HB (triangles), and RGB (squares)
321: stars per absolute visual flux unit
322: as a function of the integrated cluster magnitude ({\it top panel}) and
323: of the collision rate ({\it lower panel}). Open circles represent PCC clusters.}
324: \end{figure}
325:
326: Figure 2 shows the same results of Fig.\ 1, in a different way that
327: may be more enlightening. Here we look at the number of BSS, HB, and
328: RGB stars relative to the total flux in the same region. We call this
329: quantity $N_{\rm S}$; it is defined by
330: %
331: $$\log N_{\rm S}=\log\left[{N \over (F_{\rm HST}/10^{-0.4 V_{\rm tot}})}
332: \cdot {1 \over 10^{-0.4 M_V}}\right],$$
333: %
334: where $N$ is the total number of BSS (or HB or RGB) stars, corrected for
335: completeness, $F_{\rm HST}$ is the total flux from all the stars that we
336: measured in the region, $V_{\rm tot}$ is the integrated apparent
337: magnitude of the cluster, and $M_V$ is its integrated absolute
338: magnitude. (Note that our CMDs typically extend well below the turnoff,
339: so that the contribution from the fainter stars is negligible.)
340:
341: The first factor in the brackets can be understood as follows: the
342: quantity $F_{\rm HST}/10^{-0.4 V_{\rm tot}}$ is the fraction of the
343: total cluster flux that is sampled by the HST field. If the BSS were
344: distributed like the flux, then
345: %
346: $N / (F_{\rm HST}/10^{-0.4 V_{\rm tot}})$
347: %
348: would be the number of BSS we would expect if we could observe the whole
349: cluster.
350: In fact, as the BSS are more concentrated to the center than is the
351: flux, the quantity above is still a reasonable approximation to the
352: total number of BSS to be expected. The defect in the approximation
353: increases the scatter in $N_{\rm s}$, but it does not introduce any
354: systematic effects, since (as we have verified) there is no
355: correlation between the fraction of flux included and $M_V$.
356: The second factor in the brackets is just our previous normalization
357: to the size of the cluster, but now in luminosity units.
358:
359: Interestingly enough, Fig.\ 2 confirms that the HB and RGB stars are
360: very good tracers of the cluster population, as their absolute density
361: remains constant over more than 4 magnitudes in cluster total
362: luminosity. This fact removes the risk that the results of Fig.\ 1 might
363: be due to some anomalous gradient in the distribution of HB and RGB
364: stars (cf.\ Djorgovski, Piotto \& Capaccioli 1993). Figure\ 2 confirms
365: that the density of BSS decreases
366: with increasing total cluster mass
367: and
368: that there is no correlation between the density of BSS
369: and the collisional parameter. However, we note that, given the small
370: size of the error bars, the dispersion of the the BSS density is much
371: larger than the dispersion of the HB and RGB star densities, and that,
372: as noticed in Fig.~1, clusters with $\Gamma_\star<10^{-15}$
373: have a 2--3 times larger BSS density than clusters
374: with higher collision rates.
375: The lack of an overall dependence of $F_{\rm BSS}$ and $N_s$ on the
376: collisional parameter seems to suggest that
377: direct collisions of
378: single or binary stars
379: are
380: not the main formation mechanism of BSS.
381: At first glance, the evolution of primordial binaries also does not seem
382: to be the dominant formation mechanism for BSS in all GCs. In the
383: simple hypothesis that the binary fraction is the same in all
384: clusters, we would expect the BSS density to show a behavior similar to
385: that of the HB and RGB stars, in Fig.\ 2. On the other hand, the
386: evolution of primordial binaries is affected by the cluster
387: environment, and, in particular, it is accelerated in clusters where
388: the encounter probability is higher. Indeed, in a paper parallel to
389: this one, using the mechanism proposed by Davies \& Hansen (1998)
390: to explain the production of millisecond pulsars in GCs, Davies,
391: Piotto, \& De Angeli (2004) demonstrate that in clusters with high
392: encounter probability the formation of BSS from primordial binaries
393: has been favored in the past. Now these binaries cannot form BSS
394: anymore (they have already evolved), and this explains the observed
395: relative absence of BSS in many high mass, high collision rate
396: clusters. It also explains the relatively larger fraction of BSS
397: among the field stars, where the even lower-density environment makes
398: the evolution of binaries via encounters slower than in any GC,
399: allowing them to produce BSS for a more extended time interval
400: (till the present).
401:
402: Davies et al.\ (2004) show also that only in the most luminous GCs
403: (specifically, clusters with $M_V<-8.8$) do the BSS start to be produced
404: predominantly by stellar collisions. A better way to characterize the
405: physical properties of the BSS is to look at their luminosity function
406: (LF). In order to overcome possible dependencies of the LF on the
407: cluster metallicity, distance, and reddening, we have divided the
408: luminosity of each BSS by the turn-off luminosity of the parent cluster.
409: Figure 3 shows the LFs for GCs with different total luminosity.
410: The cut in $M_V$ has been set at $M_V=-8.8$, where the theory (Davies et
411: al.\ 2004) predicts that the BSS should become predominantly
412: collisional.
413:
414: \begin{figure}
415: \plotone{f3.ps}
416: \figcaption{BSS LFs for clusters with different integrated magnitude.}
417: \end{figure}
418:
419: Interestingly enough, clusters with $M_V<-8,8$ have a BSS LF which is
420: significantly different from the BSS LF of less luminous clusters
421: (Fig.~3), in that the LFs for the most luminous clusters have a
422: brighter peak and are significantly shifted toward brighter
423: magnitudes. If the relative importance of the BSS production
424: mechanisms depends on the cluster mass, we would then expect to see a
425: dependence of the BSS LF on $M_V$, as is observed in Fig.\ 3. In
426: general, a BSS produced by collision is expected to have a different
427: luminosity with respect to a BSS from mass transfer or merger of
428: binaries, due to the resulting interior chemical profile. How much
429: different is still controversial. Indeed, recent detailed smoothed
430: particle hydrodynamic simulations performed by Sills et al.\ (2002)
431: have shown that collision products are not chemically homogeneous.
432: This has the effect of producing a BSS structure less blue and less
433: bright than expexted on the basis of the \lq{fully mixed}\rq\ models
434: (e.g., Bailyn \& Pinsonneault 1995). Nevertheless, Sills et al.\ (2001)
435: have also shown that collision products emerge as rapidly rotating
436: blue stragglers, and so far we lack a full understanding
437: of the changes in the evolutionary properties due to rotationally
438: induced mixing.
439:
440: It is also worth noting that PCC clusters seem to have normal BSS
441: population. This might be due to the fact that the core collapse
442: phase is very short, and confined to the very central part of the
443: clusters, and therefore does not affect the BSS production over
444: the last few Gyr.
445:
446: \acknowledgments This research was supported by the Ministero
447: dell'Istruzione, Universit\`a e Ricerca (PRIN 2001 and PRIN 2002), and
448: by the Agenzia Spaziale Italiana. I.R.K.\ and S.G.D.\ acknowledge the
449: support of STScI Grants GO-6095, 7470, 8118, and 8723.
450:
451: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
452: \begin{thebibliography}{}
453: \bibitem{b95} Bailyn, C.~D. 1995, \araa, 33, 133
454: \bibitem{bp95}
455: Bailyn, C.~D., \& Pinsonneault, M.~H. 1995, \apj, 439, 705
456: \bibitem{ca01}
457: Carney, B.~W., Latham, D.~W., Laird, J.~B., Grant, C.~E., \& Morse, J.~A.
458: 2001 AJ, 122, 3419
459: \bibitem{dh98}
460: Davies, M.~B., \& Hansen, B.~M.~S. 1998, \mnras, 301, 15
461: \bibitem{dpda03}
462: Davies, M.~B., Piotto, G., \& De Angeli, F. 2004, \mnras, in press
463: \bibitem{da03}
464: De Angeli, F., et al. 2004, in preparation
465: \bibitem{dp93}
466: Djorgovski, S., Piotto, G., \& Capaccioli, M. 1993, \aj, 105, 2148
467: \bibitem{dm94}
468: Djorgovski, S.~G., \& Meylan, G. 1994, \aj, 108, 1292
469: \bibitem{d01}
470: Dolphin, A.~E. 2000, \pasp, 112, 1397
471: \bibitem{f95}
472: Ferraro, F.~R., Fusi Pecci, F., \& Bellazzini M., 1995, \aap, 294, 80
473: \bibitem{f97}
474: Ferraro, F.~R., et al. 1997, \aap, 324, 915
475: \bibitem{f03}
476: Ferraro, F.~R., Sills, A., Rood, R.~T., Paltrinieri, B., \& Buonanno, R.
477: 2003, \apj, 588, 464
478: \bibitem{fp93}
479: Fusi Pecci, F., Ferraro, F.~R., \& Cacciari, C. 1993, in Blue Stragglers,
480: ed. R.A. Saffer, ASPCS 53, 97
481: \bibitem{h96}
482: Harris, W.~E. 1996, \aj, 112, 1487
483: \bibitem{k02}
484: King, I.~R. 2002, ISBN 5-354-00163-3
485: \bibitem{MC64}
486: McCrea, W.~H. 1964, \mnras, 128, 147
487: \bibitem{mh97}
488: Meylan, G., \& Heggie, D.\ C. 1997, \aapr, 8, 1
489: \bibitem{p99}
490: Piotto, G., Zoccali, M., King, I.~R., Djorgovski, S.~G., Sosin, C.,
491: Dorman, B.,
492: Rich, R.~M., \&
493: Meylan, G.\ 1999, \aj, 117, 264
494: \bibitem{p02}
495: Piotto, G., et al. 2002, \aap, 391, 945
496: \bibitem{p00}
497: Preston, G.W., \& Sneden, C. 2000, \aj, 120, 1014 (PS00)
498: \bibitem{r03a}
499: Recio-Blanco, A., et al. 2004a, in preparation
500: \bibitem{s53}
501: Sandage, A. 1953, \aj, 58, 61
502: \bibitem{s93}
503: Sarajedini, A. 1993, in Blue Stragglers, ed.\ R.~A.\ Saffer, ASPCS 53, 14
504: \bibitem{sills02}
505: Sills, A., Adams, T., Davies, M.~B., \& Bate, M.~R. 2002, \mnras, 332, 49
506: \bibitem{z00}
507: Zoccali, M., Cassisi,
508: S., Bono, G., Piotto, G., Rich, R.~M., \& Djorgovski, S.~G.\ 2000, \apj,
509: 538,
510: 289
511: \end{thebibliography}
512:
513: \end{document}
514:
515:
516:
517:
518:
519: