1: \documentclass[apjl]{emulateapj}
2: %\documentclass[12pt,preprint]{aastex}
3:
4:
5: \shortauthors{Chandar, Fall, \& Whitmore}
6: \shorttitle{Age Distributions of Clusters and Stars}
7:
8: \def\lea{\mathrel{<\kern-1.0em\lower0.9ex\hbox{$\sim$}}}
9: \def\gea{\mathrel{>\kern-1.0em\lower0.9ex\hbox{$\sim$}}}
10: \newcommand{\lta}{{\>\rlap{\raise2pt\hbox{$<$}}\lower3pt\hbox{$\sim$}\>}}
11: \newcommand{\gta}{{\>\rlap{\raise2pt\hbox{$>$}}\lower3pt\hbox{$\sim$}\>}}
12:
13: \begin{document}
14:
15: \title{CONNECTION BETWEEN THE AGE DISTRIBUTIONS OF STAR CLUSTERS AND FIELD STARS: A FIRST APPLICATION TO THE SMALL MAGELLANIC CLOUD}
16:
17: \author{Rupali Chandar$^{1,2}$, S. Michael Fall$^{3}$, and Bradley C. Whitmore$^{3}$}
18: \affil{$^1$ Center for Astrophysical Sciences, Johns Hopkins University, 3400 N. Charles Street, Baltimore, MD 21218 \\
19: \noindent $^2$ Carnegie Observatories, 813 Santa Barbara St., Pasadena, CA 91101-1292 \\
20: \noindent $^3$ Space Telescope Science Institute, 3700 San Martin Drive,
21: Baltimore, MD 21218}
22:
23: \begin{abstract}
24:
25: We present the age distributions for star clusters and individual
26: stars in the Small Magellanic Cloud (SMC) based on data from the
27: Magellanic Clouds Photometric Survey by Zaritsky and collaborators.
28: The age distribution of the SMC clusters shows a steep decline,
29: $dN_{cluster}/d\tau \propto \tau^{-0.85\pm0.15}$, over the period $10^7
30: \lea \tau \lea 10^9$~yr. This decline is essentially
31: identical to that observed previously for more massive clusters in the
32: merging Antennae galaxies, and also for lower-mass embedded clusters
33: in the solar neighborhood. The SMC cluster age distribution therefore
34: provides additional evidence for the rapid disruption of star clusters
35: (``infant mortality''). These disrupted clusters deliver their stars
36: to the general field population, implying that the field star age
37: distribution, $dN_{fld star}/d\tau$, should have an inverse relation
38: to $dN_{cluster}/d\tau$ if most stars form initially in clusters. We
39: make specific predictions for $dN_{fldstar}/d\tau$ based on our
40: cluster disruption models, and compare them with current data
41: available for stars in the SMC. While these data do not extend to
42: sufficiently young ages for a definitive test, they are consistent
43: with a scenario wherein most SMC stars formed in clusters.
44: Future analyses of $dN_{fldstar}/d\tau$ that extend down to ages of
45: $\sim$ few million years are needed to verify the age
46: relationship between stars residing in clusters and in the field.
47:
48:
49: \end{abstract}
50:
51: \keywords{galaxies: individual (Small Magellanic Cloud) ---
52: galaxies: star clusters --- stars: formation}
53:
54:
55: \section{INTRODUCTION}
56:
57: The age distribution of a population of star clusters contains
58: important information about the formation and disruption of the
59: clusters\footnote{We use the term ``cluster'' generically to mean any
60: aggregate of stars -- regardless of mass, size, or age, and whether
61: bound or unbound -- with a density significantly higher than the local
62: stellar background, making it recognizable as a distinct entity.}.
63: The number of embedded clusters in the solar neighborhood (with masses
64: $\sim10^2-10^3~M_{\odot}$) declines rapidly with age, and it has been
65: estimated that only $4-7$\% of these clusters will survive for 100~Myr
66: (see the review by Lada \& Lada 2003). We recently found that much
67: more massive (compact) clusters ($\geq3\times10^4~M_{\odot}$) in the
68: merging ``Antennae'' galaxies also show a rapid decline in their age
69: distribution, with $dN_{cluster}/d\tau\propto\tau^{-1}$ (Fall,
70: Chandar, \& Whitmore 2005, hereafter FCW05). We interpret this
71: decline in numbers as due to a high rate of early disruption (``infant
72: mortality'') for massive clusters in the Antennae. Taken together,
73: these age distributions suggest that compact, massive clusters in the
74: chaotic environment of two merging disk galaxies disrupt at
75: approximately the same rate as lower mass, embedded clusters in the
76: more quiescent solar neighborhood. This suggests that star clusters
77: disrupt rapidly due to processes internal to the clusters themselves
78: (e.g., the removal of interstellar material due to the mass and energy
79: input from massive stars; Hills 1980; Fall 2004; FCW05), and implies
80: that the age distribution should be roughly $dN_{cluster}/d\tau
81: \propto \tau^{-1}$ in all star-forming galaxies.
82:
83: In contrast to the results for clusters in the Milky Way and in the
84: Antennae, the cluster age distribution in the Small Magellanic Cloud
85: (SMC) has been reported as $dN_{cluster}/d\tau\propto\tau^{-2.1}$
86: (Rafelski \& Zaritsky 2005; hereafter RZ05). However, this age
87: distribution is actually the number of clusters divided by the number
88: of individual stars of the same age, i.e., it is a {\it normalized}
89: cluster age distribution (RZ05). Does the proposed $\tau^{-2}$ power
90: law for the normalized cluster age distribution in the SMC differ from
91: the $\tau^{-1}$ form found in the Antennae and solar neighborhood
92: because of the normalization procedure, because clusters in the SMC
93: are forming and/or dissolving at a different rate, or for some other
94: reason? The motivation for normalizing the number of clusters by the
95: number of individual stars was to assess whether the formation of
96: stars in clusters parallels the formation of stars in the field. An
97: underlying assumption made by RZ05, then, is that star formation
98: occurs in {\it both} clusters and in the field simultaneously.
99: However, our study of the Antennae galaxies suggests that at least
100: 20\% and possibly all of the $H\alpha$ flux is emitted by compact
101: clusters, which disrupt rapidly (FCW05). This picture implies instead
102: that at least 20\% and possibly all stars form in clusters, with stars
103: being subsequently delivered from clusters to the field through the
104: process of cluster disruption. This in turn implies that
105: $dN_{cluster}/d\tau$ and $dN_{fldstar}/d\tau$ should have an inverse
106: relationship to each other.
107:
108: If the SMC cluster system is found to have an unnormalized age
109: distribution $\propto \tau^{-2.1}$, then our picture of clusters
110: disrupting at essentially the same rate in all star-forming
111: environments may have to be significantly revised. What is the
112: unnormalized SMC cluster age distribution, does it differ from the
113: normalized cluster age distribution, and what is the relationship
114: between field and cluster stars? Addressing these questions is the
115: main focus of this Letter.
116:
117: \section{AGE DISTRIBUTION OF CLUSTERS IN THE SMC}
118:
119: We use the SMC cluster properties published in RZ05 to directly determine
120: the cluster age distribution. The original cluster sample
121: was selected by Hill \& Zaritsky (2006), and we refer the reader to
122: their paper for details. Briefly, they identified cluster candidates
123: using $UBVI$ data from the Magellanic Clouds Photometric Survey (MCPS;
124: Zaritsky, Harris, \& Thompson 1997). All candidate clusters detected
125: in median-subtracted stellar images were visually inspected, and only
126: obvious clusters were retained. Clusters embedded in strong nebular
127: emission regions were excluded from the sample due to the difficulty
128: of measuring their true integrated colors. This leads
129: to a significant bias against clusters younger than $\sim10^7$~yr.
130: Due to these selection criteria and the small distance to the SMC, the
131: cluster sample of Hill \& Zaritsky (2006) is likely to be
132: approximately surface-brightness limited, rather than
133: luminosity-limited (as is typical for more distant cluster systems).
134: The adopted procedure resulted in a catalog of 204 unambiguous SMC
135: clusters.
136:
137: Basic cluster properties were estimated as follows. RZ05 measured
138: integrated $UBVI$ magnitudes within the radius enclosing 90\% of the
139: V-band light from each cluster (as tabulated in Hill \& Zaritsky
140: 2006). They then compared the colors (typically $U-B$, $B-V$, and
141: $V-I$) with the simple stellar population (SSP) model predictions of
142: STARBURST99 (Leitherer et al. 1999) and GALEV (Anders \&
143: Fritze-v. Alvensleben 2003) and adopted the age corresponding to the
144: minimum $\chi^2$. RZ05 provide electronic tables of derived cluster
145: ages and total V band luminosities for the clusters in their catalog.
146: We estimated the mass of each cluster by combining the total V band
147: luminosity provided by RZ05 with theoretical, age-dependent SSP
148: mass-to-light ratios ($M/L_V$), assuming a distance modulus of 18.87
149: (Harries et al. 2003) and a foreground reddening value of
150: $E_{B-V}=0.09$. These $M/L_V$ ratios assume a Salpeter initial
151: stellar mass function (IMF) down to $0.1~M_{\odot}$, but would be
152: $\sim1.5$ times lower if we had adopted a Chabrier-type (2003) IMF
153: that flattens below $1~M_{\odot}$. Our procedure does not account for
154: any additional extinction within the SMC, which is likely higher for
155: younger clusters, resulting in an underestimate for the masses of
156: these objects. However, the extinction in the SMC is quite small
157: along most lines of sight (e.g., Zaritsky 1999), and adopting a single
158: extinction value has little effect on our results.
159:
160: From the above data we constructed the age distribution of star
161: clusters in the SMC, $dN_{cluster}/d\tau$, by counting the number of
162: clusters ($N$) in different age bins. In this Letter, we use cluster
163: ages taken from the RZ05 comparison with the STARBURST99 model
164: (Leitherer et al. 1999) of metallicity Z=0.004, which is well-suited
165: for the young cluster population in the SMC. However, we have checked
166: that our basic results are not sensitive to the adopted metallicity,
167: population synthesis model, or bin size. The $1\sigma$ uncertainties
168: are calculated as $\pm\sqrt{N}$ for each bin. The data were
169: restricted to cover the age range over which the sample of clusters is
170: reasonably complete down to $\sim10^3~M_{\odot}$.
171:
172: \begin{figure}
173: \plotone{f1.eps}
174: \caption{
175: Age distributions of star clusters in the SMC and the Antennae.
176: The different symbols refer to
177: the approximately surface-brightness limited SMC cluster sample of RZ05
178: (filled circles), a mass-limited subset of the RZ05 cluster sample
179: (open circles), and massive star clusters in the Antennae (filled
180: triangles) as presented in FCW05. The data point for the
181: youngest bin in the SMC cluster age distributions is a lower limit, due
182: to the cluster selection procedure (see text). The solid lines
183: represent the power law $dN_{cluster}/d\tau \propto \tau^{-1}$. }
184: \end{figure}
185:
186:
187: Figure~1 shows the age distributions resulting from the full RZ05
188: sample with $\mbox{log} (\tau/\mbox{yr}) \lea 9.5$, and from the
189: approximately mass-limited sample ($\geq 10^3~M_{\odot}$). As
190: expected, the age distribution constructed from the mass-limited
191: sample is somewhat shallower than the one constructed from the entire
192: cluster sample, since there are relatively few clusters more massive
193: than $10^3~M_{\odot}$ at very young ages. Both samples have a median
194: age $\sim10^8$ yr. This is a generous upper limit on the true median
195: age due to the original selection, which deliberately excluded
196: clusters in emission-line regions, with ages $\tau \lea 10^7$~yr.
197: Regardless of this incompleteness at very young ages, the age
198: distributions of Figure~1 show that in the SMC the number of clusters
199: declines monotonically, with no obvious bends or other features. {\it
200: We find that the (unnormalized) cluster age distribution in the SMC is
201: well approximated by a power law of the form $dN_{cluster}/d\tau
202: \propto \tau^{-0.85\pm0.15}$, over the range of ages studied [$10^7 \lea
203: \tau \lea 10^9$~yr].} This form of the SMC cluster
204: age distribution is essentially the same as that found for star
205: clusters in the Antennae galaxies (FCW05). Our result therefore,
206: provides additional evidence for the rapid disruption of clusters due
207: to internal processes (``infant mortality'').
208:
209: \section{FIELD STAR AGE DISTRIBUTION AND RELATIONSHIP TO THE CLUSTERS}
210:
211: This rapid decline of $dN_{cluster}/d\tau$ with $\tau$ naturally
212: predicts that $dN_{fldstar}/d\tau$ should increase with $\tau$ as
213: dissolving clusters contribute their stars to the field. However the
214: exact form of $dN_{fldstar}/d\tau$ will depend on the fraction of
215: stars that originally form in clusters. In order to explore the
216: relationship between stars in clusters and in the field, we use the
217: statistical cluster models presented in Whitmore, Chandar, \& Fall
218: (2006; hereafter WCF06) to predict the field star age distribution.
219: Here, we provide a brief summary of the model inputs and assumptions,
220: and refer the reader to WCF06 for details. The models allow different
221: cluster formation histories, including a constant rate of formation
222: and gaussian bursts. For simplicity, we assume a constant rate of
223: cluster formation over the life of the SMC, and an initial mass
224: function of power-law form with $dN_{cluster}/dM \propto M^{-2}$ as
225: suggested by a number of recent works (e.g., Zhang \& Fall 1999;
226: Hunter et al. 2003; WCF06). After clusters form, they are disrupted
227: in two phases -- a rapid, mass-independent phase (which is modelled as
228: a power law $dN/d\tau \propto \tau^{\gamma}$, with index $\gamma$),
229: and a longer-term, mass-dependent phase that mimics the effects of
230: two-body evaporation (modelled as constant mass loss, with a rate
231: $\mu_{ev}=1\times10^{-5}~M_{\odot}~\mbox{yr}^{-1}$; see Fall \& Zhang
232: 2001 and WCF06 for details). The index of the SMC cluster age
233: distribution was found in the previous section to be
234: $\gamma=-0.85\pm0.15$. Each simulated cluster is composed of a number
235: of stars, which are assumed to be drawn from a Salpeter IMF with lower
236: and upper stellar mass limits of $0.1~M_{\odot}$ and $100~M_{\odot}$
237: respectively. The only free parameter in our model is the fraction of
238: stars, $f$, that are born initially in clusters\footnote{Our idealized
239: calculations divide stars artificially into two distinct categories,
240: i.e., ``cluster'' and ``field''. In reality, stars likely form in a
241: continuum of environments from tightly to more loosely clustered
242: aggregates.}. The models track the numbers and ages of stars in the
243: field (i.e., stars which form outside of clusters initially or end up
244: there after a cluster is disrupted), thus allowing us to predict
245: $dN_{fldstar}/d\tau$.
246:
247: Figure~2 shows our model predictions of $dN_{fldstar}/d\tau$ for three
248: different values of $f$: 0.0, 0.5, and 1.0. The (solid) curve for
249: $f=1.0$ (i.e., {\it all} star formation occurs initially in clusters) shows a
250: steep rise in the number of field stars (at very young ages), which is
251: approximately the inverse of $dN_{cluster}/d\tau$. However, by
252: $\sim10^7$~yrs the field star age distribution is essentially flat
253: (i.e., $dN_{fldstar}/d\tau \propto \tau^{0}$). This effect can be
254: understood as follows: when $dN_{cluster}/d\tau$ is plotted
255: logarithmically, the {\it rate} at which the cluster population
256: diminishes is observed as the slope, $\gamma$. However, if
257: $dN_{cluster}/d\tau \propto \tau^{-0.85}$ were plotted linearly rather
258: than logarithmically, one would see that the number of clusters
259: declines quickly at first, and then begins to flatten out. The field
260: star age distribution is roughly the inverse of the cluster age
261: distribution, and so must necessarily rise very quickly as stars are
262: dispersed into the field. This increasing function essentially
263: ``saturates'' by $\sim10^7$~yrs, predicting that $dN_{fldstar}/d\tau$
264: varies rapidly only for a short time.
265:
266: \begin{figure}
267: \plotone{f2.eps}
268: \caption{Comparison between the SMC field star age distribution
269: (stars) from HZ04 and predictions for $dN_{fldstar}/d\tau$ from the
270: WCF06 models (see text for details). The power-law index $\gamma$,
271: which quantifies the rate of cluster disruption, is fixed at $-0.85$,
272: as measured from the SMC cluster age distribution. The only free
273: parameter in the model is the fraction of stars, $f$,
274: that form within clusters. Note that even if
275: 100\% of stars form initially in clusters, the field star distribution
276: flattens by $\sim10^7$~yrs. }
277: \end{figure}
278:
279:
280: Is the field star age distribution in the SMC consistent with most stars
281: forming in clusters? Harris \& Zaritsky (2004; hereafter HZ04)
282: present the star formation history of the SMC. They use all detected
283: stars (i.e., in both clusters and in the field), from the MCPS
284: ground-based data. However, there is a natural bias in their stellar
285: catalog against stars in clusters, due to their inability to
286: photometer in dense regions. Therefore, we refer to stars in their
287: sample as belonging to the ``field''. We determine the number of SMC
288: stars as a function of age by dividing the star formation rates
289: presented in Table~2 of HZ04 by the mean stellar mass. Figure~2 shows
290: a comparison of the field star age distribution in the SMC with our model
291: predictions. This shows that over the observed range the field star
292: age distribution is essentially flat (although see HZ04 for a detailed
293: analysis of small-scale variations in the star formation history).
294: Unfortunately, the youngest age in the HZ04 compilation is just where
295: the predicted rise in the field star age distribution begins to
296: flatten, making it impossible to determine $f$ with this data set.
297: Additionally, the HZ04 data do not discriminate between stars in
298: clusters and stars in the field, which results in ``pollution'' of
299: $dN_{fldstar}/d\tau$ from cluster stars. Nevertheless, Figure~2 shows
300: that despite $dN_{fldstar}/d\tau$ being rather flat, {\it the SMC
301: field star data do not contradict our scenario where most star
302: formation occurs in short-lived clusters}. Our simulations
303: establish that even if 100\% of stars formed originally in clusters,
304: their rapid rate of disruption creates the field star population so
305: quickly that after $\sim10^7$~yrs the two populations will always
306: appear to evolve independently.
307:
308:
309: \section{DISCUSSION AND CONCLUSIONS}
310:
311: We have shown here that the (unnormalized) cluster age distribution in
312: the SMC has the form $dN_{cluster}/d\tau \propto \tau^{-0.85\pm0.15}$
313: and that overall the field star age distribution is nearly flat (i.e.,
314: $dN_{fldstar}/d\tau\propto \tau^{0}$, over the age range $10^7 \lea
315: \tau \lea 10^9$~yr). The {\it normalized} SMC
316: cluster age distribution (the number of clusters divided by the number
317: of individual stars of the same age) therefore must have the same form
318: as the unnormalized distribution [i.e., $(dN_{cluster}/d\tau)_{norm}
319: \propto \tau^{-0.85\pm0.15}$]. Our conclusion contradicts that of
320: RZ05, who found that the normalized cluster age distribution is
321: $(dN_{cluster}/d\tau)_{norm} \propto \tau^{-2.1}$ from the same data.
322: The difference in the power-law exponent between our work and that of
323: RZ05 comes about because we divided the cluster age distribution
324: ($dN_{cluster}/d\tau$) directly by that for the field stars
325: ($dN_{fldstar}/d\tau$), whereas RZ05 divided
326: $dN_{cluster}/d\mbox{log}\tau$ by $dN_{fldstar}/d\mbox{log}\tau$, and
327: then appear to have introduced an extraneous factor of $\tau^{-1}$.
328:
329: Our result that the age distribution of SMC clusters has a power-law
330: form $dN_{cluster}/d\tau \propto \tau^{-0.85\pm0.15}$ has important
331: implications regarding the physical processes that disrupt star
332: clusters. The fact that clusters in vastly different environments
333: (including the chaotic merging of two galaxies, a quiescent spiral
334: disk, and a gas-rich dwarf galaxy) have age distributions with nearly
335: identical forms provides strong evidence that star clusters disrupt
336: rapidly due to processes {\it internal} to the clusters themselves.
337: Over the first few million years photoionization, stellar winds, jets
338: and supernovae can remove much of the interstellar matter from
339: clusters leaving the member stars freely expanding (e.g., Hills
340: 1980; Boily \& Kroupa 2003a,b). These expanding clusters will be
341: observable for $\sim$~few~$\times 10^{7}$~yrs after becoming unbound,
342: until their surface brightness becomes so low that they become
343: indistinguishable from statistical fluctuations of stars in the field
344: (FCW05). In addition to the removal of interstellar material, mass
345: loss from stars themselves will likely contribute to the continued
346: unbinding of clusters (e.g., Applegate 1986; Chernoff \& Weinberg
347: 1990; Fukushige \& Heggie 1995). Finally, two-body evaporation will
348: destroy any remaining clusters on longer timescales (e.g., Vesperini
349: 1997; Baumgardt 1998; Fall \& Zhang 2001). The early disruption
350: (``infant mortality'') appears to operate relatively independent of
351: cluster mass, while two-body evaporation is mass-dependent.
352:
353: The stars from disrupted clusters add to the general field, requiring
354: a close relationship between the age distributions of clusters and
355: field stars if most stars form initially in clusters. In this work we
356: compared age distributions for clusters and individual stars in the
357: SMC with model predictions to further test this general framework.
358: Our models suggest that the rapid drop in the cluster age distribution
359: should be reflected in a rapid rise in the field star age
360: distribution. However this rapid rise in the number of field stars is
361: only observable for the first $\sim10^7$~yrs {\it regardless of the
362: fraction of stars that initially formed in clusters}. Because the
363: present data on SMC stars begin just as the predicted field star
364: distributions flatten, and because there is no discrimination between
365: stars belonging to clusters versus to the field, they do not provide
366: an unambiguous answer. However the observed relations are consistent
367: with a scenario where most stars form in clusters that are
368: subsequently disrupted by the mechanisms discussed above. The models
369: also predict that even if all stars form initially in clusters,
370: galaxies will contain a relatively small number of observable clusters
371: at any given time. Future studies with the {\it Hubble Space
372: Telescope} (which allow better discrimination between stars in
373: clusters and in the general field), and in the infrared (which can
374: probe very young, deeply embedded stars and clusters) could be used to
375: search for the expected rise in the field star age distribution at
376: $\tau \lea 10^7$~yrs. A measurement of the rise in
377: $dN_{fldstar}/d\tau$ would provide a direct estimate of the fraction
378: of stars which originally form in clusters versus in the field.
379:
380: \acknowledgments
381:
382: We thank Jason Harris for explaining the HZ04 table of SMC star
383: formation rates, Dennis Zaritsky for discussions about the SMC cluster
384: age distribution, and Mike Santos and Francois Schweizer for helpful
385: comments on the manuscript. R. C. is grateful for support from NASA
386: through grant GO-09192-01-A from STScI, which is operated by AURA,
387: Inc., under NASA contract NAS5-26555.
388:
389:
390: \begin{thebibliography}{}
391:
392: \bibitem{anders03} Anders, P. \& Fritze-v. Alvensleben, U. 2003, A\& A, 401, 1063
393:
394: \bibitem{applegate86} Applegate, J. H. 1986, \apj, 301, 132
395:
396: \bibitem{baumgardt98} Baumgardt, H. 1998, A \& A, 330, 480
397:
398: \bibitem{boily03} Boily, C. M. \& Kroupa, P. 2003a, MNRAS, 338, 665
399:
400: \bibitem{boily03} Boily, C. M. \& Kroupa, P. 2003b, MNRAS, 338, 673
401:
402: \bibitem{bc03} Bruzual, G., \& Charlot, S. 2003, MNRAS, 344, 1000
403:
404: \bibitem{chabrier03} Chabrier, G. 2003, PASP, 115, 763
405:
406: \bibitem{chernoff90} Chernoff, D. F. \& Weinberg, M. D. 1990, ApJ, 351, 121
407:
408: \bibitem{fall04} Fall, S. M 2004, in ASP Conf. Ser. 322, The Formation and Evolution of Massive Young Star Clusters, ed. H. J. G. L. M. Lamers, L. J. Smith, \& A. Nota (San Francisco: ASP), 399
409:
410: \bibitem{fall05} Fall, S. M., Chandar, R., \& Whitmore, B. C. 2005, ApJ, 631, L133 (FCW05)
411:
412: \bibitem{fall01} Fall, S. M., \& Zhang, Q. 2001, ApJ, 561, 751
413:
414: \bibitem{fukushige95} Fukushige, T. \& Heggie, D. C. 1995, MNRAS, 276, 206
415:
416: \bibitem{harries03} Harries, T. J., Hilditch, R. W., \& Howarth, I. D. 2003, MNRAS, 339, 157
417:
418: \bibitem{hz04} Harris, J. \& Zaritsky, D. 2004, AJ, 127, 1531 (HZ04)
419:
420: \bibitem{hz03} Hill, A. \& Zaritsky, D. 2006, AJ, 131, 414
421:
422: \bibitem{hills80} Hills, J. G. 1980, ApJ, 225, 986
423:
424: \bibitem{hunter03} Hunter, D., Elmegreen, B. G., Dupuy, T. J., \& Mortonson, M. 2003, \aj, 126, 1836
425:
426: \bibitem{lada03} Lada, C. J. \& Lada, E. L. 2003, ARA \& A, 41, 57
427:
428: \bibitem{sb99} Leitherer, C. et al. 1999, ApJS, 123, 3
429:
430: \bibitem{miller79} Miller, G. \& Scalo, J. 1979, ApJS, 41, 513
431:
432: \bibitem{rz05} Rafelski, M. \& Zaritsky, D. 2005, AJ, 129, 2701 (RZ05)
433:
434: \bibitem{vesperini97} Vesperini, E. 1997, MNRAS, 287, 915
435:
436: \bibitem{whitmore06} Whitmore, B. C., Chandar, R., \& Fall, S. M. 2006, submitted to AJ (WCF06)
437:
438: \bibitem{zaritsky99} Zaritsky, D. 1999, \aj, 118, 2824
439:
440: \bibitem{zaritsky97} Zaritsky, D., Harris, J., \& Thompson, I. 1997, AJ, 115, 1002
441:
442: \bibitem{zhang99} Zhang, Q., \& Fall, S. M. 1999, \apjl, 527, 81
443:
444: \end{thebibliography}
445:
446:
447:
448:
449:
450: \end{document}
451: