6 Conclusions

The main results and their implications presented in this work are the following:

1.

On the basis of high resolution ground-based images, taken with VLT and FORS1, I identified 71 bona fide globular clusters in the halo of the nearest giant elliptical galaxy NGC 5128. Only 5 of these clusters were previously known;

2.

In a UV color-magnitude diagram for the complete sample of 71 clusters and a UVK color-color diagram for a subsample of 23 clusters, the objects span a similar magnitude and color range as the globular clusters in the MW and M 31. Note, however, that there is no cluster corresponding to $\omega$ Cen, the brightest MW cluster, in these two fields. The position of 23 objects in the (U-V)₀ vs. (V-K)₀color-color diagram is consistent with their classification as old globular clusters;

3.

The GCLFs spanning -10.1<M_V<-4.9and -9.3<M_U<-3.3 have been constructed. These are the deepest GCLFs of an elliptical galaxy made so far. Kolmogorov-Smirnov statistics show that the difference between the GCLFs of NGC 5128 and MW is not larger than the difference between the GCLFs of M 31 and MW. Similarity of the GCLFs of an elliptical with respect to the spiral galaxy had never before been tested at the faint end;

4.

The presence of faint globular clusters in the halo of NGC 5128 puts constraints on the effectiveness of the tidal forces in the deep elliptical galaxy potential. The dynamical effects may be important for clusters that are found within $\sim$2 $R_{\rm e}$ from the galactic center. Unfortunately, the selection of globular clusters from VLT images is not sensitive to the most compact and the faintest clusters, similar to ones like Pal 1, Pal 13, AM 4 and Terzan 1 in our Galaxy, while the less dense ones start to be confused with background galaxies at faint magnitudes;

5.

The (U-V)₀ color histogram of 71 clusters indicates a bimodal distribution, supporting the Zepf & Ashman  (1993) suggestion. Assuming that the clusters in NGC 5128 and in the MW span a similar age interval, and adopting the linear fit between the (U-V)₀color and metallicity, I derived the [Fe/H] of the red and blue peaks of the bimodal distribution to be -1.7 dex and -0.6 dex, respectively. Using the SSP models from Kurth et al. (1999), instead of the linear fit, the results do not change significantly, giving values of -1.7 dex and -0.5 dex for the metal-poor and metal-rich peaks, respectively. This is different from the Harris et al. (1992) and Zepf & Ashman (1993) result, but is partially due to the adopted color cut of ( U-V)₀<2.5 in selecting globular clusters and to small number statistics, since both samples have <10% of the total cluster population. Relaxing the red color cut, the mean color of the distribution corresponds to a mean metallicity of [Fe/H]=-0.7 dex, very similar to the Harris et al. (1992) result. The true nature of the redder objects has to be assessed through spectroscopy.

Acknowledgements

I am grateful to Dante Minniti for his guidance and help with this project. I thank Andres Meza for providing the code that calculates the kernel estimator. I would like to acknowledge the referee for valuable comments which led to improvements in the paper and useful discussions with Markus Kissler-Patig, Thomas Puzia and Steve Zepf. Thanks also go to Tim Bedding, Dave Silva and Elena Pancino. This research was partially supported by FONDECYT grant No. 1990440.