7 Summary and conclusions

We obtained near-infrared H-band profile decompositions for 75 galaxies taken primarily among dwarf galaxies in the Virgo cluster. Adding these new observations to the ones similarly taken in the Virgo, Coma and A1367 clusters and in the "Great Wall'' (Paper V), we gathered H-band data for 818 galaxies. These include all galaxies brighter than $m{\rm _p} =15.7$ in the Coma region, corresponding to $M_{\rm p}<-19.2$ ($\mu=34.9$) and 94% of galaxies brighter than $m\rm _p =14.0$ in the Virgo cluster, corresponding to $M_{\rm p}<-17.2$ ($\mu=31.2$), thus the observations of giant galaxies are complete. Considering only the Virgo cluster, we also covered 30% of galaxies in the interval $14.0<m{\rm _p}<16.0$ corresponding to $-17.2<M_{\rm p}<-15.2$, thus to the transition region between giant and dwarf galaxies (see Sandage et al. 1985). The completenes in the same magnitude range increases from 30 to 47% if one considers the ISO sample only. The studied sample is representative of all Hubble types, including dE and Im, and spans 4 orders of magnitude in luminosity.

We model the surface brightness profiles of the studied galaxies with either a de Vaucouleurs r^1/4 law (D), an exponential law (E), a combination of the two (M), or with a profile that is truncated at the periphery (T). Using the fitted quantities we find that:
1) Less than 50% of the giant elliptical galaxies have pure D profiles;
The majority of giant galaxies (E to Sb) is best represented by a M profile. Scd-BCD galaxies have pure exponential profiles;
2) Most dwarf galaxies (independently from their detailed morphological type) follow exponential profiles or truncated decompositions;
3) The type of decomposition is a strong function of the total H band luminosity ( $10^8<L_{\rm H}<10^{11.5}$ $L_\odot$), irrespective of the galaxy Hubble classification: the fraction of pure exponential profiles decreases with increasing luminosity, while that of M ones increases with luminosity. Truncated profiles are characteristic of the lowest luminosity galaxies. Pure D profiles are absent at low luminosities $L_{\rm H}<10^{10}$ $L_\odot$ and become dominant above 10¹¹ $L_\odot$;
4) The light concentration index C₃₁ (presence of central cusps and extended outer haloes) is a strong non-linear function of the total luminosity, irrespective of the Hubble classification: dwarf systems have low C₃₁, typical of exponential disks; high C₃₁, characteristic of conspicuous bulges, are found only at the highest luminosities. There exist however a class of bulge-less, high luminosity galaxies. These are giant Sc's;
5) dE galaxies have mildly redder colors and higher C₃₁ than dIs. The only subclass of early-type dwarfs having structural parameters indistinguishable from those of late-type dwarfs seems to be that of dE-pec, which therefore represents the possible missing link between dEs and dIs. This is supported by the evidence of post-starburst activity found in the dE-pec VCC1499.
The results summarized in points 1) through 4) should not suffer from selection biases since at low-luminosities, where our sample is severely incomplete, we observed primarily the highest surface brightness galaxies. Thus a bias, if any is present, should select in favour of high C₃₁ galaxies with D or M profiles, because at comparable luminosities these objects have higher central surface brightness than those with low C₃₁ and E or T profiles.
Summarizing, points 1-4 indicate that the frequency of occurrence of relevant cusps and extendend luminous haloes, absent among low-mass galaxies, increases significantly with increasing mass. This is consistent with the monolithic collapse scenario (Sandage 1986) provided that the collapse efficiency scales with mass (Gavazzi & Scodeggio 1996). If, otherwise, merging is invoked as the mechanism for building galaxies of increasing mass, a problem arises: while extended haloes are naturally produced as remnants of mergers between stellar disks, central high-brightness cusps require that the mergers occur in the presence of a gaseous phase (Hernquist et al. 1993). If this were the case, however, cusps (bulges) would be composed of younger stellar populations than it is generally observed.

Acknowledgements

We thank C. Bonfanti for the reduction and analysis of OHP spectra of two galaxies