2 Observations and globular cluster selection

CCD images of 15 fields centered on low surface brightness (LSB) galaxies in the Fornax Cluster were obtained (with the original purpose of studying the LSB galaxies, Cellone et al. 1994; Cellone & Forte 1996) with the 0.90-m and 1.50-m telescopes at CTIO (Chile), during two observing runs in October 1989 and November 1990, and using the C and T₁ filters of the Washington photometric system. The dwarf galaxies are listed in Table 1: the first column gives their FCC numbers (Ferguson 1989) and the second one gives the respective angular distances from NGC 1399, which will be considered as the center of the Cluster. The dwarfs are distributed in angular distance from 12$^\prime$ up to 175$^\prime$ from NGC 1399 (see Fig. 1); as we will adopt a distance modulus of 31.35 for the Fornax Cluster throughout this paper (Madore et al. 1999), that corresponds to projected distances ranging between 65 and 950 kpc from the cluster center. The fields sizes range from 10 to 17 arcmin²; hence, they have the advantage of being larger than the ones studied by Miller et al. (1998a) to search for cluster candidates. However, they are not as deep as the HST images: we identify cluster candidates up to $T_1 \approx$ 22 mag while Miller et al. reach $V \approx$ 25 mag (equivalent to $T_1 \approx$ 24.5 mag, according to the relation between these magnitudes obtained from Geisler 1996 and Cellone et al. 1994). For a detailed description of the observations we refer to Cellone et al. (1994).

Figure 1: $\;$DSS image of the Fornax Cluster with NGC 1399 at the center. The squares and the FCC numbers identify the 15 dwarf galaxies. The segment corresponds to 1$^\circ$. North is up and East to the left.

In order to identify the globular cluster candidates, we selected the point sources within certain ranges of colors and magnitudes. The analysis of each frame was carried out in the following steps:

1.

The dwarf galaxies were first removed using the IRAF / STSDAS tasks "ellipse" and "bmodel", which allowed us to fit elliptical isophotes to the galaxies, create a model based on them and subtract the model from the image.

2.

Using the photometric software SExtractor (Source-Extractor, Bertin & Arnouts 1996) we detected all the sources in the dwarf-subtracted frames, we measured their instrumental magnitudes and colors, and each object was classified with an "stellarity-index", a tool for performing a reliable resolved/unresolved source separation. Among the different types of magnitudes that SExtractor measures, the "corrected isophotal'' were selected as the most reliable ones for stellar-like sources (Arnouts et al. 1997).

3.

The instrumental magnitudes and colors were transformed to the standard system via the equations derived by Cellone et al. (1994), which were applied to the new measurements because they were both obtained by means of aperture photometry. The reddening towards the dwarfs' fields was obtained from the Schlegel et al. (1998) maps and transformed into the Washington system by means of the relations given by Harris & Canterna (1979). As the mean estimated color-excess $E_{C-T_{1}} \approx$ 0.02 mag is much smaller than the photometric errors (see below), it was considered negligible.

  

Table 1: Dwarf galaxies and surface densities of globular cluster candidates ( $\delta _{\rm GC}$) in their fields.

4.

All the sources classified with an "stellarity-index" below 0.35 were considered as resolved sources and were discarded. The limiting index was estimated running DAOPHOT on several images as well as adding artificial stars to the frames, and comparing the results for different values of the "stellarity-index" with the results for the DAOPHOT indices "sharpness", "chi" and "roundness" that allow the identification of point sources, galaxies and image defects.

5.

Finally, we selected as globular cluster candidates those point sources with: magnitudes 19 < T₁ < 22 mag and colors 0.8 < (C-T₁) < 2.2 mag. We estimate that the completeness of the sample is larger than 90% above magnitude $T_1 \approx$ 22 mag. This conservative cutoff prevents against any possible bias introduced by small variations in the photometric limit between different fields. Reassuringly, we found no correlation between effective exposure (considering telescope diameter and integration time) and surface density of globular cluster candidates. The number of globular cluster candidates that was identified in each dwarf galaxy's field ranges from zero to ten, which corresponds to surface densities between 0.00 and 0.71 candidates/arcmin², as is listed in the third column of Table 1.

Figure 2: Color-magnitude diagram for all the detected sources (crosses) and the globular cluster candidates (open circles). Dotted lines represent the limits of the globular cluster selection.

Figure 3: (C-T1) color distribution: the dashed line corresponds to the raw data for the globular cluster candidates, the dot-dashed line to the comparison field and the solid line shows the result after the background subtraction. Dotted lines represent the limits of the globular cluster selection. The metallicity scale (Geisler & Forte 1990) is given on top.

Figure 2 shows the color-magnitude diagram of all the identified sources (about 350), and the ones selected as globular cluster candidates (75). The mean photometric errors for the globular cluster candidates are 0.09 mag in T₁ and 0.15 mag in (C-T₁).