Rebecca A. W. Elson and Basilio X. Santiago
Institute of Astronomy, Madingley Road, Cambridge CB3 0HA,
England
globular cluster
candidates in the Virgo galaxy M87. The sample is essentially
complete to V = 26.
We find a strongly bimodal color
distribution with
peaks at 
and 1.23.
A comparison with the color distribution of globular clusters in the
Milky Way shows that M87 contains a large population of red clusters
that is virtually absent in our Galaxy. We also
find that the luminosity functions of the red and blue clusters
are different. The brighter clusters are predominantly blue, while the
fainter clusters are roughly equally divided between red and blue.
The presence of two distinct sub-populations with
different mean luminosities
casts doubt on the common assumption of a
universal luminosity function for globular clusters.
Increasingly accurate measurements of the colors and metallicities of globular clusters in distant galaxies are beginning to shed new light on, and provoke new questions about, the origin of globular cluster systems. Color and, by inference, metallicity distributions are being revealed as bimodal or even multi-modal, and appear to vary significantly from one host/parent galaxy to the next (cf. Ajhar, Blakeslee & Tonry 1994). This has prompted the suggestion that distinct sub-systems of clusters formed in distinct episodes, perhaps during mergers (cf. Ashman & Zepf 1992). In this contribution we examine a field in M87 (NGC 4486), the central giant elliptical galaxy in the Virgo cluster, observed with HST as part of the Medium Deep Survey Key Project (cf. Griffiths et al. 1994). A full report of the results may be found in Elson & Santiago (1996).
Our study is based on a field located 
arcmin from the center
of M87, obtained with HST's Wide Field Camera (WFC-2) on 1995 August 5.
The F814W (
Cousins I) and F606W (
Cousins V) filters
were used. Two V images were obtained, with total exposure time of
2600 seconds, and four I images, with total exposure time 4700 seconds.
Objects were detected automatically and measured using DAOPHOT.
Instrumental HST magnitudes were transformed to the Johnson-Cousins
system using the calibrations from Holtzman et al. (1995). Globular
cluster candidates were selected on the basis of color and morphology.
We corrected for contamination from compact, spherical background
galaxies using a deep field from the Medium Deep Survey database.
Our final sample contains 
clusters with 
.
Figure 1 shows a color distribution of the M87 globular clusters. Two
narrow peaks are evident; one at 
, and one at 
.
Fitting two Gaussian functions to the bimodal distribution for clusters
with 
, yields 
mag for both the red and
blue peaks; we find the same 
for only the brightest clusters,
with 
. This value is several times greater than the Poisson
errors which at 
are 
. There is, therefore,
a small but real spread in the intrinsic colors of both the red and
blue clusters.
Figure: A comparison of the color distributions of the
M87 clusters (
)
(N=254), with the color distribution of 91 Milky
Way globular clusters from Cameron Reed et al. (1988).
The M87 histogram has been
divided by the ratio of sample sizes. It is clearly bimodal.
The large population of
red clusters evident in M87 is almost entirely absent in our Galaxy.
Figure 1 also shows such a comparison, between the M87 clusters with
from both our samples, and 91 Milky Way globulars from
Cameron Reed, Hesser, & Shawl (1988). Our histogram has been divided
by the ratio of the sample sizes. The Galactic clusters have 
which, at the distance modulus of M87, corresponds to 
.
Thus, the magnitude limits of the two samples are similar. There are
relatively fewer blue clusters in M87, and the substantial population of
red globulars present in M87 is almost entirely absent in the Milky Way.
Figure 2 shows histograms of color for the full sample divided into
sub-samples of bright and faint clusters. The bright clusters (
) are clearly dominated by blue objects. 67% have 
,
while only 23% have 
. For the fainter clusters (
) the fractions are 56% (blue) and 44% (red).
Figure: Histogram of colors for faint (N=126)
and bright (N=128) clusters in the combined sample. The bright sample
is clearly dominated by blue clusters, while the faint sample has
similar numbers of red and blue clusters.
Our results suggest the presence of two distinct populations of globular
clusters in M87, each with an intrinsic color spread, which probably
originated during two distinct episodes, or in two different modes of
cluster formation. The color distribution of the blue population in M87
is virtually indistinguishable from that of the globular clusters in
our Galaxy. The red population on the other hand, is almost entirely
absent in our Galaxy. One possible interpretation of this is that the
color reflects metallicity, and that there is a dearth of metal rich
globulars in our Galaxy. According to Couture et al. (1990), the red
colors would imply approximately solar metallicity. They would, therefore,
have formed out of gas that had already been significantly enriched.
Indeed, the metallicity of M87 itself is [Fe/H]
(Brodie
& Huchra 1991), consistent with the inferred metallicity of the
population of red clusters. If elliptical galaxies formed from the
merger of spiral galaxies (cf. Toomre 1977), then the blue population
may represent clusters native to the galaxies which merged, while the
red population may have formed during the merger. Indeed, several
galaxies currently undergoing mergers have recently been observed to
contain blue point-like objects, which may be young globular clusters
(cf. Holtzman et al. 1992, Whitmore et al. 1993).
An alternative interpretation of the color differences is that they
reflect a difference in age, in the sense that the bluer clusters
are younger. This seems unlikely, however, given the similarity in
color between the blue M87 clusters and those in the Galaxy, which are
known to be old. Certainly the blue clusters are not blue enough to
be suggestive of a recent episode of cluster formation in M87. The 40
bright point-like objects observed in NGC 7252 by Whitmore et al. (1993)
have mean color 
, considerably bluer than the blue
population in M87, which peaks at 
.
The red clusters in our sample seem to be on average fainter than the
blue clusters. This does not necessarily imply a difference in mass,
however. In a metal poor globular cluster the giants, which dominate the
integrated light, would be brighter than those in a metal rich cluster
by as much as two magmitudes in 
for a change in metallicity from
[Fe/H] 
to 
(cf. Bergbusch & VandenBerg 1992).
The blue clusters may, therefore, only appear brighter because they are more
metal poor. It is interesting that Whitmore et al. (1995) do not find a
luminosity difference between the red and blue clusters in their sample.
This may reflect a difference in the population of clusters near the
center, and further out in M87.
In any case, the existence of two distinct populations of clusters with differing mean luminosity casts doubt on the popular concept of a universal luminosity function for globular clusters. And even if such a thing existed, measuring the peak luminosity accurately at these distances is very difficult, even with HST. Indeed, now that globular cluster color distributions have clearly been shown to vary from galaxy to galaxy, any detailed coincidence among the corresponding luminosity functions must be attributed to a conspiracy between the distributions of cluster masses and metallicities. This is particularly true between spirals, which appear to contain only blue globular clusters, and ellipticals, which may contain both red and blue clusters. It will be interesting to see whether increasingly accurate photometry reveals differences among luminosity functions, as suggested by our results, and indeed by Ajhar et al. (1994) who find some suggestion of a non-universal luminosity function for the one galaxy in their sample with clusters intermediate between red and blue. It would also be interesting to obtain spectra of a sample of red and blue clusters to determine whether their metallicities are indeed different, and whether there is any evidence for kinematic differences between the two populations.
Ajhar, E. A., Blakeslee, J. P., & Tonry, J. L. 1994, AJ 108, 2087
Ashman, K. M. & Zepf, S. E. 1992, ApJ 384, 50
Bergbusch, P. & VandenBerg, D. 1992, ApJS 81, 163
Brodie, J. & Huchra, J. 1991, ApJ 379, 157
Cameron Reed, B., Hesser, J., & Shawl, S. 1988, PASP 100, 545
Couture, J., Harris, W. E., & Allwright, J. W. B. 1990, ApJS 73, 671
Elson, R. & Santiago, B. 1996, MNRAS, accepted
Griffiths, R., et al. 1994, ApJ 437, 67
Holtzman, J. A., et al. 1992, AJ 103, 691
Holtzman, J. A., et al. 1995, preprint
Toomre, A. 1977, in The Evolution of Galaxies and Stellar Populations, ed. B. M. Tinsley & R. B. Larson (Yale University Observatory, New Haven), p. 401
Whitmore, B., Schweizer, F., Leitherer, C., Borne, K., & Robert, C. 1993, AJ 106, 1354
Whitmore, B., Sparks, W., Lucas, R., Macchetto, F., & Biretta, J. 1995, ApJ 454, L73