Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stsci.edu/~marel/abstracts/psdir/n4449pap_3.ps Äàòà èçìåíåíèÿ: Mon Dec 4 20:07:11 2000 Äàòà èíäåêñèðîâàíèÿ: Sat Dec 22 13:13:20 2007 Êîäèðîâêà: IBM-866 Ïîèñêîâûå ñëîâà: http astrokuban.info astrokuban

A Young Stellar Cluster in the Nucleus of NGC 4449
Torsten BØoker 1 , Roeland P. van der Marel
Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218
boeker@stsci.edu, marel@stsci.edu
Lisa Mazzuca 2
Goddard Space Flight Center, Greenbelt, MD 20771
lmazzuca@pop500.gsfc.nasa.gov
HansíWalter Rix, Gregory Rudnick
MaxíPlanckíInstitut fØur Astronomie, KØonigstuhl 17, Dí69117 Heidelberg, Germany
rix@mpiaíhd.mpg.de, rudnick@mpiaíhd.mpg.de
Luis C. Ho
Observatories of the Carnegie Institution of Washington, 813 Santa Barbara Street, Pasadena,
CA 91101í1292
lho@ociw.edu
Joseph C. Shields
Ohio University, Department of Physics and Astronomy, Clippinger Research Laboratories, 251B,
Athens, OH 45701í2979
shields@helios.phy.ohiou.edu
ABSTRACT
We have obtained 1í2 Ú A resolution optical Echellette spectra of the nuclear star
cluster in the nearby starburst galaxy NGC 4449. The light is clearly dominated by
a very young (6 - 10 Myr) population of stars. For our age dating, we have used
recent population synthesis models to interpret the observed equivalent width of stellar
absorption features such as the H I Balmer series and the Ca II triplet around 8500 Ú A.
We also compare the observed spectrum of the nuclear cluster to synthesized spectra
1 A#liated with the Astrophysics Division, Space Science Department, European Space Agency.
2 Current address: Department of Astronomy, University of Maryland, College Park, MD 20742

-- 2 --
for stellar populations of varying ages. All these approaches yield a consistent cluster
age. Metallicity estimates based on the relative intensities of various ionization lines
yield no evidence for significant enrichment in the center of this low mass galaxy: the
metallicity of the nuclear cluster is about one fourth of the solar value, in agreement
with independent estimates for the disk material of NGC 4449.
Subject headings: galaxies: individual (NGC 4449) --- galaxies: starburst --- galaxies:
nuclei.
1. Introduction
Recent observations with the Hubble Space Telescope (HST) at both optical (Phillips et al.
1996; Matthews et al. 1999) and nearíinfrared (NIR, Carollo, Stiavelli, & Mack 1998; BØoker et
al. 1999a) wavelengths have revealed that a compact, photometrically distinct star cluster is often
present in the center of spiral galaxies of all Hubble types. The formation mechanism of such
clusters is puzzling, especially in lateítype spirals which by definition do not possess a prominent
bulge. In these cases, the gravitational potential is very shallow, and it is far from obvious how the
gas could fall all the way to the center to form a dense star cluster.
In this paper, we describe the results of a pilot program to investigate the processes of nuclear
cluster formation. More specifically, we have performed 1í2 Ú A resolution optical Echellette specí
troscopy of the nucleus of NGC 4449, which is part of a small sample of galaxies with prominent
nuclear star clusters that were identified from a recent NIR survey with HST (BØoker et al. 1999a).
The morphological classification of NGC 4449 has been the subject of some confusion. It was origí
inally classified as Irr by Hubble (1936) and Sandage (1961), but it is listed as Sm in Sandage
& Tamman (1981). As van den Bergh (1995) points out, the presence of a stellar nucleus might
be a useful criterion to distinguish between the two classes, since Irregulars usually do not have a
wellídefined nucleus í in contrast to lateítype spirals, as discussed above. The extent, morphology,
and dynamics of its H I gas suggest that NGC 4449 has undergone some interaction in the past, a
fact that might well account for its unusually high star formation activity (Hunter, van Woerden
& Gallagher 1999). It also has a prominent stellar bar which covers a large fraction of the optically
visible galaxy. Both the nuclear cluster and the stellar bar are evident in Figure 1, which shows a
groundíbased Ríband image by Frei et al. (1996).
The fact that the nucleus of NGC 4449 is also prominently seen in the far ultraviolet (FUV)
(Hill et al. 1998) and in various emission lines (Sabbadin et al. 1984; Mayya 1994; Hill et al. 1998;
BØoker et al. 1999a) indicates that it is the site of ongoing star formation. NGC 4449 is somewhat
unusual in this regard, since emissioníline surveys have shown that the nuclei of lateítype galaxies
often are devoid of ionized gas (e.g. BØoker et al. 1999a). It appears that NGC 4449 is one of the
few lateítype spirals that is ``caught in the act'' of ongoing star formation in its nucleus.
NGC 4449 was chosen for a case study because of its high clusterítoídisk contrast which allows
us to isolate the light of the nuclear cluster from the underlying galaxy. From the diskísubtracted

-- 3 --
cluster spectrum, we derive the equivalent width (EW) of a number of stellar absorption features.
We then use stellar population synthesis methods to constrain the star formation history of the
nuclear cluster and to determine the age of its dominant stellar population. In ç 2, we describe
our observations and the data reduction methods. Section 3 describes the quantitative analysis of
a number of emission lines and stellar absorption features, as well as our spectral synthesis e#orts.
From the results of these various diagnostic tools, we deduce an age of about 6 - 10 Myr for the
nuclear cluster. In ç 4, we discuss our results and compare them to a recent study by Gelatt,
Hunter, & Gallagher (2001) which confirms our age estimate from a complementary dataset. We
also briefly discuss suggested mechanisms of nuclear cluster formation and their possible impact on
the evolution of the host galaxy.
2. Observations and data reduction
The data described here were taken in April 1999 at the Steward Observatory 90 ## Bok reflector.
We used the Boller & Chivens spectrograph in Echellette mode with a 2 ## íwide slit to obtain spectra
with 1í2 Ú A resolution covering the range # 3600 - 9000 Ú A. The average seeing over the observation
was 2.5 ## .
Following Willmarth & Barnes (1994), we used standard IRAF 3 routines in the LONGSLIT
and ECHELLE packages to perform the following data reduction steps. After the data frames
were biasísubtracted and trimmed using the CCDPROC routine, we corrected for bad pixels. We
obtained a bad pixel mask by ratioing two domeflat exposures with di#erent integration times, and
replaced the bad pixels with a median filter. A second pass through CCDPROC then flatfielded
the data using a medianícombined set of domeflat exposures.
We obtained three 30 minute exposures of NGC 4449 which were medianícombined after verí
ifying that the brightness profile was properly centered on the slit for all three exposures. Using
the APALL task, we then extracted the multiíorder spectrum of the galaxy over 12 subapertures
of 1 ## each, thus retaining the spatial information. Spectra of the standard star Hiltner600 were
extracted with the DOECSLIT task and used to measure the Echelette sensitivity function, based
on spectrophotometry published by Massey et al. (1988). Also within DOECSLIT, we determined
the dispersion correction from spectra of a HeliumíArgon lamp. Both flux and wavelength calibraí
tion were then applied to the galaxy spectrum, before we used SCOMBINE to combine all orders
into one twoídimensional spectrum covering the full wavelength range.
Finally, we averaged the central 2 ## into a oneídimensional spectrum of the nucleus, as well
as two stripes of 3 ## width each, located 3.5 ## on either side of the nucleus. Subtracting the two
resulting 1íd spectra from each other eliminates the sky emission lines and isolates the light of
the nuclear star cluster from the underlying galaxy disk. The simple linear interpolation of the
underlying disk is likely to leave a small residual contribution to the spectrum of the nuclear cluster.
3 IRAF is distributed by the National Optical Astronomical Observatories, which are operated by AURA, Inc.
under contract to the NSF.

-- 4 --
Higher order interpolation clearly is desirable, but proved unfeasible over the full wavelength range
of the spectrum. However, for limited wavelength ranges, we verified that the results of our analysis
are robust against such variations in the methodology of the background subtraction. We conclude
that the errors introduced by the imperfect interpolation of the galaxy surface brightness profile
are small.
As the result of the described data reduction process, the fluxí and wavelengthícalibrated
spectrum between 3600 and 9000 Ú A of the central 2 ## of NGC 4449 is shown in Figure 2. The red
end of the spectrum is subject to some residual fringing with peakítoípeak amplitudes smaller
than 10% of the continuum. This slightly a#ects the uncertainty of the the continuum level for
measurement of the Ca triplet absorption lines (see ç 3.1.2). The large residuals near 7580 Ú A are
due to imperfect correction of telluric oxygen absorption (``Aíband''), as are smaller residuals at
6280 Ú A and 6860 Ú A.
3. Results
3.1. Stellar Absorption Features
In order to measure the equivalent width (EW) of various stellar absorption features, we
rectified and normalized the spectrum. The line and continuum windows that we used for continuum
fitting and EW measurement are defined in GonzÒalez Delgado, Leitherer & Heckman (1999) for
the Balmer and HeI series, and in DÒÐaz, Terlevich & Terlevich (1989) for the CaII triplet (CaT)
around 8500 Ú A. They are listed in Table 1, together with all measured EWs. Figure 3 shows the
rectified spectrum over two exemplary regions which contain some higher order Balmer lines and
the CaT, respectively. The analysis of the Balmer absorption is complicated by the presence of
emission lines. We took the following approach to subtract the line emission before measuring
the EW of the absorption features. We iteratively subtracted Gaussian components (typically one
or two) until the resulting ``pure'' absorption spectrum of each Balmer line could be well fitted
by a Voigtíprofile. Figure 4 demonstrates the process. The sum of the fitted Gauss components
then yields the emission line fluxes for the Balmer series. Fitting synthesized spectra of stellar
populations as described in ç 3.2 yields an alternative approach. After subtraction of the best
fitting model spectrum, the residuals contain only the line emission. The line fluxes derived with
both methods are consistent. We conservatively estimate the uncertainty in the derived EWs of
the Balmer absorption to be 10%. In what follows, we compare the measured EWs to evolutionary
synthesis models.
3.1.1. Balmer absorption
In Figure 5, we compare our measurements to evolutionary synthesis models of GonzÒalez
Delgado et al. (1999). Each panel contains the time evolution of the EW of a specific absorption

-- 5 --
feature. The grey horizontal bar denotes the respective measured EW, the width of the bar indicates
the uncertainty of 10%. The crossing point of the bar with the model yields a solution for the cluster
age. All features yield an age of 6-10Myr for the dominant stellar population in the nuclear cluster
of NGC 4449. This result is rather independent of the detailed IMF or the star formation mode
(continuous or instantaneous). For all reasonable models, the rise in the Balmer EW occurs at about
the same time, so that the crossing point of the measurements with the model curve is robust to
within a few million years. A theoretical second solution, namely an extremely old cluster with an
age around 10 Gyr is ruled out from the evidence that follows.
3.1.2. Calcium triplet
Because the CaT absorption is produced in cooler and more evolved stars than the Balmer
series, it provides an independent age estimator. In Figure 3b, we show the spectral region comí
prising the CaII triplet around 8500 Ú A which also contains the line and continuum windows used
for our analysis. These are also listed in Table 1 and are identical to those used by GarcÒÐaíVargas,
MollÒa & Bressan (1998) to derive theoretical EWs of the CaT from evolutionary synthesis models.
We compare the EW sum of the two strongest features (#8542 and #8662) as measured from
our data to the GarcÒÐaíVargas et al. (1998) models (their grid I) in Figure 6. The horizontal line
again denotes the measured EW of the CaT and the shaded region its estimated uncertainty which
is dominated by the uncertainty in the continuum level due to residual fringing on the red end of
the spectrum. Despite the conservative error estimate, the very deep CaT absorption in the nuclear
cluster of NGC 4449 unambiguously confirms a young cluster age between 5 and 20 Myr.
Figure 6 also seems to indicate a high metallicity of at least solar value. However, we caution
that the lowímetallicity stellar evolutionary tracks used in the population synthesis models have
been shown to systematically underpredict the red supergiant (RSG) features (Origlia et al. 1999).
This is a longstanding problem that also manifests itself in the inability of current models to
reproduce the observed blueítoíred supergiant ratio in galaxies (e.g. Langer & Maeder 1995).
As a consequence, the predicted CaT EW for lowímetallicity RSG populations falls significantly
below the observed values in, e.g., young clusters in the LMC (Bica, Santos, & Alloin 1990). For
individual stars, the CaT EW exceeds 9 Ú Aonly in red supergiants (DÒÐaz et al. 1989). Certainly,
any singleíage population of stars that has a CaT EW as high as the nuclear cluster of NGC 4449
has to contain red supergiants with ages between 5 and 20 Myr, irrespective of its metallicity. In
summary, Figure 6 confirms a young cluster age, but does not provide a reliable way to constrain
its metallicity. In ç 3.3.2, we will explore alternative methods to derive the metallicity based on
nebular emission lines.
3.2. Spectral modeling
As an additional test of the above results we have compared the spectrum of the nuclear
cluster in NGC 4449 to spectral models, using the entire available wavelength range. For this we

-- 6 --
first created model spectra of simple stellar populations (SSPs), i.e. populations that form in
instantaneous, shortílived starbursts. Models were calculated for ages on a logarithmically spaced
grid with a spacing of 0.5 dex. We used the 1996 version of the Bruzual & Charlot (1993) models,
which assume a Salpeter (1955) IMF between 0.1 and 125 M# and solar metallicity. The observed
spectrum is then modeled as a linear superposition of these SSPs, which essentially allows an
arbitrary star formation history (approximated as a weighted sum of delta functions). We find the
weights that optimize the fit to the data, allowing also for a redshift due to the systemic velocity of
the galaxy. The # 2 fit was performed with the software of Rix & White (1992), developed originally
for the analysis of galaxy kinematics.
The best fit to the observed spectrum was obtained with a model that has the large majority
of its light (> 90%) in the SSP of age 10 7 years. The dataímodel comparison is shown in Figure 7.
The overall shape of the spectrum as well as the depth of the Balmer absorption features are well
reproduced. However, the Ca II feature at 3934 Ú A, as well as some other metal lines, are significantly
underestimated in the fit. This discrepancy is possibly due to a metallicity mismatch. We used
template spectra with solar abundances for the modeling because we did not have lower metallicity
tracks with su#cient spectral resolution available. However, as we will show in ç 3.3.2, the nuclear
cluster in NGC 4449 does actually have subsolar metallicity. Nonetheless, the shape of the 4000 Ú A
break and the depth of the Balmer series absorption features depends only slightly on metallicity,
so that the quality of the fit should yield a good age estimate despite the noníperfect choice of
metallicity. We conclude that the spectral modeling confirms the results of our previous analysis
that the nuclear cluster in NGC 4449 is young.
3.3. Emission lines
As can be seen from Figure 2, the spectrum of the nuclear star cluster of NGC 4449 -- after subí
traction of the circumínuclear light -- contains a number of strong emission lines. After subtracting
the continuum fit described in ç 3.2, we measured the fluxes of all significant emission lines (except
H9 and H10) using SPECFIT as implemented in IRAF (Kriss 1994). The residual spectrum was
subdivided into three segments and a fit with # 2 optimization was performed, assuming a linear
fit to any residual continuum and Gaussian profiles in a common velocity frame for the emission
lines. The fluxes of H9 and H10 were measured as described in ç 3.1 to more accurately correct for
the underlying absorption. The resulting fluxes on an absolute scale and relative to H# are listed
in Table 2, along with their 1í# uncertainties.
3.3.1. Extinction
Extinction of the nebular emission can be quantified from reddening of the spectrum as iní
dicated by the Balmer emission line ratios. For the intrinsic relative line strengths, we employed
predictions from Hummer & Storey (1987) assuming Case B recombination in a 10 4 K gas with

-- 7 --
n e = 100 cm -3 (e.g., H#/H# = 2.86). The electron density can be constrained from the ratios of
[O II] # 3729/# 3726 = 1.38 ‘ 0.02 and [S II] # 6716/# 6731 = 1.29 ‘ 0.05, both of which suggest
n e # 100 cm -3 (e.g. Osterbrock 1989). For the extinction analysis, we employed the reddening
law from Cardelli, Clayton & Mathis (1989) with R V = 3.1. The measured Balmer emission fluxes
ratioed to H# then imply extinction values of AV =(1.1, 0.85, 1.4, 1.8, 0.47, 2.6) mag for (H#, H#,
H#, H#, H8, H9). The measured flux for H8 is almost certainly contaminated by emission in He I
#3890, which would account for the low value of AV suggested by this line. The other features
show a general trend of increasing AV with higheríorder Balmer lines, which is probably an indicaí
tion that the line fluxes are still slightly underestimated due to stellar absorption. We ultimately
adopted an extinction value of AV = 1.1 mag based on the H#/H# ratio, since among the Balmer
lines these features have the highest signalítoínoise ratio and are expected to su#er the least from
underlying absorption. The majority of this extinction is evidently intrinsic to NGC 4449, since
the foreground Galactic extinction is only 0.06 mag (Schlegel, Finkbeiner, & Davis 1998). The AV
adopted here is about one magnitude higher than the value derived by Ho, Filippenko & Sargent
(1997) using the same intrinsic line ratio. On the other hand, Ho et al. (1997) removed the stellar
continuum by subtracting generic template spectra in order to measure the emissioníline strengths.
The di#erence of about 30% in the measured line ratio needed to reconcile the two measurements
can probably be explained by any one or a combination of the following reasons: a slight mismatch
of the continuum template in the Ho et al. (1997) analysis, the fact that they used a somewhat
larger aperture, or di#erences in spectral resolution (which a#ect the ability to separate the emission
core from the absorption trough).
3.3.2. Metallicity
The relative line fluxes can be used to obtain an independant estimate for the metallicity in
the nucleus of NGC 4449. The ``brightíline'' method originally proposed by Pagel et al. (1979)
makes use of the quantity
R 23 # (I 3727 + I 4959 + I 5007 )/H # (1)
as a measure of metallicity. For the NGC 4449 nucleus, the reddeningícorrected line fluxes imply
R 23 = 8.2 ‘ 0.2. A recent analysis illustrating the available calibrations of R 23 is provided by
Kobulnicky, Kennicutt, & Pizagno (1999). Their Figure 8 indicates that our measured value of
R 23 falls at or near the turnaround point between the lowí and highímetallicity branches for this
diagnostic; this result implies 12 + log(O/H) # 8.3 - 8.4, or about one fourth of the solar value
(12 + log(O/H) = 8.9). This abundance for the nucleus is in good agreement with the average
value for the disk of NGC 4449 of 8.32 reported by Lisenfeld & Ferrara (1998). Abundances can
be inferred directly if diagnostics of the nebular electron temperature are available. Table 2 lists
a detection of the auroral [O III] #4363 line, which in combination with [O III]##4959, 5007 can
be used for this purpose. We used the dereddened measurements to calculate the O/H abundance
using the methods described in Oey & Shields (2000). The large error bar on the [O III] #4363 flux
translates into substantial uncertainties in the derived quantities, with a resulting temperature of
16700 ‘ 3500 K and 12 + log(O/H) = 7.8 ‘ 0.3. The latter abundance is consistent at the 2# level
with the result based on the R 23 diagnostic.

-- 8 --
3.4. Mass and color of the nuclear cluster
Assuming a cluster age of 10 Myr, a simple estimate of the cluster mass can be obtained
from its Víband luminosity. Our fluxícalibrated spectrum after subtraction of the underlying
galaxy light (Figure 2) yields F 550nm # 3.5 ½ 10 -15 ergs s -1 cm -2 Ú A -1 , or m V = 15.0. This
value can be compared to evolutionary synthesis models for singleíage populations such as those
in the Starburst99 package (Leitherer et al. 1999). For a Salpeter IMF with m low = 1 M# and
m up = 100M# , the observed cluster luminosity requires a mass of about 4½10 5 M# (after correction
for AV = 1.1, for a distance 4 to NGC 4449 of 3.9 Mpc and an age of 10 Myr). This cluster mass
should represent a lower limit for two reasons. First, the distance to NGC 4449 is rather uncertain
and has been estimated as high as 5.4 Mpc (KraaníKorteweg & Tamman 1979). In addition, the
IMF likely has a lower mass cuto# below the assumed 1M# . Adding lowímass stars does not change
the cluster luminosity much, but significantly increases the total mass.
As an additional check of our results, we have measured the color of the nuclear cluster. The
(V - I) color of the spectrum in Figure 2 is (V - I) = 1.1. Correcting for an extinction of AV = 1.1
as derived from the nebular emission lines (ç 3.3.1), the intrinsic color of the stellar population
is (V - I) = 0.6. This value is most likely a lower limit, since at UVí and optical wavelengths,
the stellar continuum is usually less reddened than the nebular emission lines (Calzetti 1997).
Comparison with the Leitherer et al. (1999) models shows that a color of (V - I) # 0.6 is entirely
consistent with a 6 - 10 Myr old population.
The mass in the nuclear cluster implied by the stellar continuum can be compared with that
inferred from the nebular flux. The extinctionícorrected H# flux of 8.9 ½ 10 -14 ergs s -1 cm -2
corresponds to a total H# luminosity of 1.6 ½ 10 38 ergs s -1 . For Case B recombination (ç 3.3.1),
this luminosity requires that hydrogeníionizing photons be produced and absorbed at a rate of
Q(H) = 3.4 ½ 10 50 s -1 within the H II region. Comparison to the Starburst99 models shows that
this is entirely consistent with a 10 Myr old cluster with a mass of 4 ½ 10 5 M# and a metallicity of
about one fourth of the solar value.
The dereddened He I #5876/H# ratio of 0.10‘0.01 can be compared with the Case B prediction
(Benjamin et al. 1999) of 0.14( He + /H +
0.1
), where He + /H + is the ionic abundance ratio. The fact that
the observed line ratio is close to the prediction implies that most of the helium within the nuclear
H II region is singly ionized. This condition sets a strong limit on the age of the cluster, since
relatively hot stars are required to maintain the helium ionization. The production rate of heliumí
ionizing photons implied by the extinctionícorrected He I #5876 flux is 2.6 ½ 10 49 s -1 . For the
above cluster parameters, this luminosity is consistent with the Starburst99 prediction at an age
of # 7 Myr.
4 derived by assuming that NGC 4449 follows the Hubble flow with H0 = 65 km s -1 Mpc -1

-- 9 --
4. Discussion and outlook
The results of this study imply that NGC 4449 can be added to a growing list of lateítype spiral
galaxies that host a compact, young star cluster in their nucleus. While accurate spectroscopy í
and hence a reliable age estimate í of nuclear clusters is available for only a few objects such as the
Milky Way (Krabbe et al. 1995), NGC 7793 (Shields & Filippenko 1992), and IC 342 (BØoker, van
der Marel & Vacca 1999b), photometric measurements show that many clusters are fairly blue, e.g.
those in M 31, M 33, NGC 4242, or ESO 359í029 (Lauer et al. 1998; Matthews et al. 1999). This
indicates the existence of a young (# 100 Myr) stellar population in many of these nuclei.
From an HST survey of intermediateítype spirals, Carollo et al. (1998) find that virtually
all galaxies with exponential bulges (i.e. bulges with an exponential surface brightness profile, as
opposed to ``powerílaw'' or ``R 1/4 '' bulges) host compact nuclear clusters and that their luminosity
correlates with that of the host galaxy. If one assumes that the massítoílight ratio of the host
galaxies as a whole is roughly the same for all objects, the most straightforward interpretation of
this result is that more massive galaxies have more massive nuclear clusters. On the other hand,
stellar populations fade with time (e.g. Bruzual & Charlot 1993). Another possible interpretation
therefore is that more massive galaxies host younger clusters. This intrinsic degeneracy makes age
dating of nuclear clusters from photometric data alone impossible. Spectroscopic analysis is needed
to investigate the age distribution of nuclear clusters and to decide whether their formation is a
one time event or not. If nuclear clusters indeed form recurrently, the formation processes likely
impact the morphological evolution of the host galaxy.
Of particular interest in this context is a model suggested by Friedli & Benz (1993) and
developed further by Norman, Sellwood & Hasan (1996). These authors point out that buildíup
of a central mass concentration í e.g. a nuclear star cluster í can dissolve stellar bars, and lead to
the formation of a bulge via collective bending instabilities (Raha et al. 1991; Merritt & Sellwood
1994). The Norman et al. (1996) simulations show that only 5% of the combined disk and bar
mass in a central concentration is su#cient to destroy a stellar bar on short timescales, leading to
a bulgeílike distribution of stars. In this scenario, repeated cycles of bar formation --- gas infall
--- cluster formation --- bar disruption will build the galaxy bulge over time. If true, the Hubble
classification scheme could be naturally explained as an evolutionary sequence from late to early
Hubble types.
In the case of NGC 4449, the cluster mass of (at least) 4 ½ 10 5 M# derived in ç 3.4 would
be su#cient to destroy a bar with a mass up to 10 7 M# . The unusually prominent stellar bar in
NGC 4449, which covers a large part of the optically visible galaxy, is probably much more massive
than this value. This can be seen from the total luminosity of NGC 4449, which is listed in the
RC3 (de Vaucouleurs et al. 1991) as V 0
T = 9.53, which corresponds to LV # 2 ½ 10 9 L#,V . Since the
optical image is dominated by the stellar bar, and typical massítoílight ratios are of order unity,
the bar mass should be # 10 9 M# . The fact that it coíexists with the nuclear cluster is therefore
not inconsistent with the Norman et al. (1996) scenario.
However, as pointed out before, NGC 4449 is a special case because of its unusually high overall
star formation rate and its complex history with evidence for past interactions. The complex gas

-- 10 --
flows produced by a past merger might well explain why -- di#erent from most other lateítype
spirals (e.g. BØoker et al. 1999a) -- the nucleus of NGC 4449 contains large amount of ionized gas.
Furthermore, there is no indication of an underlying older population of stars in the nuclear cluster
of NGC 4449, as is the case, e.g., in the Milky Way. We point out, however, that for external
galaxies where individual stars cannot be resolved, it is di#cult to detect an old stellar population
underlying a young starburst because the latter is dominating the luminosity (assuming similar
masses for both bursts). It is somewhat unclear at this point whether the processes that govern
the star formation in the nucleus of NGC 4449 are similar to those in ``typical'' lateítype spirals, or
whether NGC 4449 is an entirely di#erent ``special'' case.
After the present paper had been finished, another study of the nuclear cluster in NGC 4449
appeared in the literature (Gelatt et al. 2001). Using groundíbased spectroscopy of the 12 CO(2, 0)
and 12 CO(3, 1) absorption features around 2.3²m and UVI colors from HST imaging, these authors
derived a cluster age between 8 and 15Myr, in good agreement with our estimate of 6-10Myr. Así
suming a somewhat lower extinction to the NGC 4449 nucleus than derived by us from the Balmer
decrement (ç 3.3.1), Gelatt et al. conclude that the color of the nuclear cluster is better matched
by evolutionary tracks with a metallicity of Z=0.008, somewhat higher than the average over the
NGC 4449 disk. However, increasing the extinction by one magnitude (roughly the di#erence beí
tween our extinction and theirs) would easily move the location of the nuclear cluster in their
Figure 3 within the range of the lower metallicity tracks. After accounting for the di#erent extincí
tion estimates, the derived VíI color of the nuclear cluster di#ers by about 0.2 magnitudes in the
two studies. This could be the result of the somewhat di#erent aperture sizes used to extract the
cluster light (1.4 ## for Gelatt et al. compared to 2.0 ## for our study). Either way, both papers come
to very similar conclusions on the age of the nuclear cluster in NGC 4449, with complementary
methods.
This study is part of a larger observational program that contains three independent, but
related approaches. First, we will use the WFPC2 onboard the Hubble Space Telescope during
Cycle 10 to measure the structural properties of all nuclear clusters found in a large sample of
nearby, faceíon spiral galaxies of late Hubble type. We will then measure the stellar velocity
dispersion in the brightest nuclei found in order to derive the cluster mass, as recently demonstrated
for the example of IC 342 (BØoker et al. 1999b). Finally, we hope to obtain optical spectra with
HST/STIS to derive the stellar population and to constrain the formation history of individual
nuclear clusters in much the same way as described in this paper. We aim to study a large number
of nuclear clusters to allow a statistical analysis of their properties and dependence on the Hubble
type of the host galaxy. Since groundíbased observations are possible only for the brightest nuclear
clusters, they are likely biased towards the young end of their age distribution. The goal of our study
is to eliminate this bias. Measuring the true age distribution of nuclear star clusters will allow us
to answer the question whether repeated nuclear starbursts are indeed common in lateítype spirals,
and, if so, what their duty cycle is.

-- 11 --
5. Summary
We have presented a new, highíquality spectrum of the nuclear star cluster of NGC 4449 which
covers the region between 3700 and 9000 Ú A. Analysis of a variety of stellar absorption features
leads to the conclusion that the nucleus of NGC 4449 has undergone a shortílived starburst about
6 - 10 Myr ago. We do not find evidence for significant metal enrichment due to multiple previous
periods of star formation in the nucleus of NGC 4449.
We thank the anonymous referee for useful comments that helped to improve the presentation
of this paper.

-- 12 --
REFERENCES
Athanassoula, E. 1992a, MNRAS, 259, 328
Athanassoula, E. 1992b, MNRAS, 259, 345
Benjamin, R. A., Skillman, E. D., & Smits, D. P. 1999, ApJ, 514, 307
Bica, E., Santos, J. F. C., Jr., & Alloin, D. 1990, A&A, 235, 103
BØoker, T., et al. 1999a, ApJS, 124, 95
BØoker, T., van der Marel, R. P., & Vacca, W. D. 1999b, AJ, 118, 831
Bruzual, A. & Charlot, S. 1993, ApJ, 405, 538
Calzetti, D. 1997, AJ, 113, 162
Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245
Carollo, C. M., Stiavelli, M., & Mack, J. 1998, AJ, 116, 68
Courteau, S., de Jong, R., & Broeils, A. 1996, ApJL, 457, L73
de Vaucouleurs, G. 1948, Ann. d'Astrophysique, 11, 247
de Vaucouleurs, G., de Vaucouleurs, A., Corwin, H., Buta, R. J., Paturel, G., & Fouque, P. 1991,
Third Reference Catalogue of Bright Galaxies, New York:SpringeríVerlag
DÒÐaz, A. I., Terlevich, E., & Terlevich, R. 1989, MNRAS, 239, 325
Frei, Z., Guhathakurta, P., Gunn, J. E., & Tyson, J. A. 1996, AJ, 111, 174
Friedli, D. & Benz, W. 1993, A&A, 268, 65
GarcÒÐaíVargas, M. L., MollÒa, M., & Bressan, A. 1998, A&AS, 130, 513
Gelatt, A. E., Hunter, D. A., & Gallagher, J. S. 2001, PASP, in press (February issue)
GonzÒalez Delgado, R. M., Leitherer, C., & Heckman, T. M. 1999, ApJS, 125, 489
Hill, R. S. et al. 1998, ApJ, 507, 179
Ho, L. C., Filippenko, A. V., & Sargent, W. L. W. 1997, ApJS, 112, 315
Hubble, E. 1936, The Realm of the Nebula, Yale University Press, New Haven
Hummer, D. G., & Storey, P. J. 1987, MNRAS, 224, 801
Hunter, D. A., van Woerden, H., & Gallagher, J. S. 1999, AJ, 118, 2184
Kobulnicky, H. A., Kennicutt, R. C. Jr., & Pizagno, J. L. 1999, ApJ, 514, 544

-- 13 --
Kormendy. J. & Bender, R. 1996, ApJ, 464, 119
KraaníKorteweg, R. C. & Tamman, G. A. 1979, Astron. Nachr. 300, 181
Krabbe, A. et al. 1995, ApJ, 447, L95
Kriss, G. 1994, in ASP Conf. Ser. 61, Astronomical Data Analaysis Software and Systems III, ed.
D. R. Crabtree, R. J. Hanisch, & J. Barnes (San Francisco: ASP), 437
Lauer, T. R., Faber, S. M., Ajhar, E. A., Grillmair, C. J., & Scowen, P. A. 1998, AJ, 116, 2263
Langer, N. & Maeder, A. 1995, A&A 295, 685
Leitherer, C., et al. 1999, ApJS, 123, 3
Lisenfeld, U. & Ferrara, A. 1998, ApJ, 496, 145
Massey, P., Strobel, K., Barnes, J. V., & Anderson, E. 1988, ApJ, 328, 315
Matthews, L. D. et al. 1999, AJ, 118, 208
Mayya, Y. D. 1994, AJ, 108, 1276
Merritt, D. & Sellwood, J. A. 1994, ApJ, 425, 551
Norman, C. A., Sellwood, J. A., Hasan, H. 1996, ApJ, 462, 114
Oey, M. S., & Shields, J. C. 2000, ApJ, 539, 687
Origlia, L., Goldader, J. D., Leitherer, C., Schaerer, D., & Oliva, E. 1999, ApJ, 514, 96
Osterbrock, D. E. 1989, Astrophysics of Gaseous Nebulae and Active Galactic Nuclei, University
Science Books:Mint Valley, CA
Pagel, B. E. J., Edmunds, M. G., Blackwell, D. E., Chun, M. S., & Smith, G. 1979, MNRAS, 189,
95
Phillips, A.C., Illingworth, G. D., MacKenty, J. W., & Franx, M. 1996, AJ, 111, 1566
Raha, A., Sellwood, J.A., James, R., & Kahn, F.D. 1991, Nat, 352, 411
Rix, H.íW. & White, S. D. M. 1992, MNRAS, 254, 389
Sabbadin, F., Ortolani, S., & Bianchini, A. 1984, A&A, 131, 1
Salpeter, E. E. 1955, ApJ, 121, 161
Sandage, A. 1961, The Hubble Atlas of Galaxies, Carnegie Institution, Washington
Sandage, A. & Tamman, G. A. 1981, A Revised ShapleyíAmes Catalog of Bright Galaxies, Carnegie
Institution, Washington

-- 14 --
Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525
Shields, J. C. & Filippenko, A. V. 1992, ASP Conf. series, 31, 267
van den Bergh, S. 1995, AJ, 110, 613
Willmarth, D. & Barnes, J. 1994, A User's Guide to Reducing Echelle Spectra with IRAF, National
Optical Astronomy Observatories, Arizona
This preprint was prepared with the AAS L A T E X macros v5.0.

-- 15 --
Fig. 1.--- Groundíbased Ríband image of NGC 4449 from Frei et al. (1996). The field is 7 # ½ 7 # ,
with North up and East to the left. The greyscale has been optimized to the dynamic range of the
galaxy. The nuclear cluster is clearly visible.
Fig. 2.--- Top: spectrum of the nuclear star cluster of NGC 4449, taken over a 2 ## ½2 ## aperture. The
underlying disk/bulge emission has been subtracted. Bottom: the same spectrum on an expanded
yíscale. The residuals near 7580 Ú A, 6280 Ú A, and 6860 Ú A are due to imperfect correction of telluric
oxygen absorption (``Aíband''). A few sky lines have not been perfectly corrected, and appear as
negative ``spikes''.
Fig. 3.--- Normalized and rectified spectrum of the Balmer series (top) and the CaT absorption
features (bottom). The horizontal bars denote the line and continuum windows that were used for
measurement of the EW (Table 1).
Fig. 4.--- Spectrum of H# before (top) and after (bottom) removal of the line emission. The Gauss
and Voigt profile fits that were used to subtract the emission line and to judge the completeness of
the line removal are overplotted in the respective panels (dashed lines).
Fig. 5.--- Comparison of the Balmer EW with evolutionary synthesis models of GonzÒalez Delgado
et al. (1999) for an instantaneous burst with a Salpeter (1955) IMF between M low = 1 M# and
M up = 80 M# . The horizontal grey bar indicates the measured EW for the nuclear cluster in
NGC 4449, its width corresponds to the measurement uncertainty. The crossing point of bar and
model indicates an age of # 10 Myr.
Fig. 6.--- Comparison of the CaT EW with evolutionary synthesis models of GarcÒÐaíVargas et
al. (1998) for an instantaneous burst with a Salpeter (1955) IMF between M low = 0.8 M# and
M up = 100 M# . The horizontal grey bar indicates the measured EW for the nuclear cluster in
NGC 4449, its width corresponds to the measurement uncertainty. The crossing point of bar and
model indicates an age of # 10 Myr. The lowímetallicity models are highly uncertain, the fact that
they do not reproduce the observed EW should not be considered as evidence for a high cluster
metallicity (see discussion in text).
Fig. 7.--- Results of the spectral modeling. The modeled spectrum for a 10 Myr stellar population
with a Salpeter (1955) IMF, based on the Bruzual & Charlot (1993) evolutionary tracks (thick
line) is plotted over the observed spectrum of the nuclear cluster in NGC 4449. The spectrum has
a resolution of # 1 Ú A, about 5 times higher than the model.

-- 16 --
Table 1. Absorption features
Species Line window EW [ Ú A] Cont. window
H10 #3799 3783í3813 2.81 a
H9 #3835 3823í3853 3.72 a
H8 #3889 3874í3904 3.96 a
H# #3970 3955í3985 2.01 a
H# #4102 4070í4130 5.03 b
H# #4340 4311í4371 3.42 c
H# #4862 4830í4890 3.88 d
Ca II #8498 8483í8513 4.05 e
Ca II #8542 8527í8557 6.78 e
Ca II #8662 8647í8677 4.45 e
Note. --- Continuum windows are defined as follows:
(a) 3740:3743,3760:3762,3782:3785,3811:3812,3909:3915
(b) 4019:4020,4037:4038,4060:4061,4138:4140,4148:4150
(c) 4301:4305,4310:4312,4316:4318,4377:4381,4392:4394,4397:4398
(d) 4820:4830,4890:4900
(e) 8447í8462,8842í8857
Windows (a)í(d) are taken from GonzÒalez Delgado et al.
(1999), (e) is from DÒÐaz et al. (1989)

-- 17 --
Table 2. Emission Line Fluxes
Species F [10 -15 ergs/s/cm 2 ] F/F(H#) F c /F c (H#)
[O II] #3726 42.3‘0.6 1.56‘0.03 2.27‘0.05
[O II] #3729 58.4‘0.6 2.15‘0.04 3.13‘0.06
H10 #3798 #0.2 #0.01 í
H9 #3836 0.9‘0.2 0.03‘0.01 0.04‘0.02
[Ne III] #3869 4.2‘0.5 0.15‘0.02 0.22‘0.03
H8 #3889 2.5‘0.4 0.09‘0.01 0.13‘0.02
H# #3970 2.6‘0.3 0.1‘0.01 0.13‘0.02
H# #4102 5.0‘0.4 0.18‘0.02 0.24‘0.02
H# #4340 11.0‘0.3 0.41‘0.01 0.48‘0.01
[O III] #4363 1.2‘0.4 0.04‘0.01 0.05‘0.02
H# #4862 27.1‘0.4 1.00‘0.02 1.00‘0.02
[O III] #4959 20.0‘0.4 0.74‘0.02 0.72‘0.02
[O III] #5007 58.9‘0.4 2.17‘0.04 2.04‘0.03
[He I] #5876 3.3‘0.2 0.12‘0.01 0.10‘0.01
[O I] #6300 2.9‘0.4 0.11‘0.01 0.08‘0.01
[N II] #6548 4.2‘0.4 0.15‘0.01 0.11‘0.01
H# #6563 109.5‘0.3 4.04‘0.06 2.82‘0.04
[N II] #6583 13.5‘0.3 0.5‘0.01 0.35‘0.01
[S II] #6716 13.0‘0.3 0.48‘0.01 0.32‘0.01
[S II] #6731 10.1‘0.3 0.37‘0.01 0.25‘0.01
[Ar III] #7136 3.1‘0.3 0.11‘0.01 0.7‘0.01
Note. --- Column 2: Line flux as measured from Figure 1. Colí
umn 3: measured line ratio relative to H#. Column 4: line ratio
relative to H# after correction for an extinction of AV = 1.1 mag.
Uncertainties listed are 1#.

-- 18 --

-- 19 --

-- 20 --

-- 21 --
4070 4080 4090 4100 4110 4120 4130
4.0
.7
.8
4.5
.9
5.0
5.5
1
6.0
Flux
(10eí15
erg/cm/s/A)
Wavelength (angstroms)
normalized
flux
6.5

-- 22 --

-- 23 --

-- 24 --