Astron. Astrophys. 345, 419-429 (1999)

6. Summary and concluding remarks

Adopting the medium resolution spectral library obtained by J84 and the SSP models by Bressan et al. (1994, 1996) and Tantalo et al. (1996) we have calculated three narrow band indices in the spectral range 4200 Å, namely H+K(CaII), H/FeI and 4000, as a function of age and metallicity.

With the aid of these results, we simulated post-star-burst galaxies by means of the simple recipe: an old host galaxy (with age of 15 Gyr and solar composition) to which at some arbitrary age a burst of star formation is added. The intensity of this is measured by the percentage of mass with respect to the total which is turned into stars by the burst episode.

We have shown that the line strength indices H+K(CaII), H/FeI and 4000 respond to even small traces of past star formation activity. In particular, we have discussed the sensitivity of H+K(CaII) to the ongoing star formation events due to the contamination of the Balmer H emission line.

From the analysis of the location of shell and pair galaxies on the diagnostic planes based on the indices H+K(CaII), H/FeI and 4000 we may advance some suggestions, whose validity is, however, of mere statistical nature owing to the large uncertainty still affecting the data.

1. Shell galaxies and early-type galaxies members of pairs in the diagrams show different distributions in the 4000 vs. H+K(CaII) and H/FeI vs. H+K(CaII) planes.

2. In shell galaxies, the age of the last star forming event goes from 0.1 to several Gyr and involves different percentages of mass. If the last burst of stellar activity that affects the line strength indices, correlates with the dynamical mechanism, i.e. merger or weak interaction, forming the shell features, in such a case these latter are long lasting phenomena .

3. The distribution of early-type galaxies members of pairs suggests that the vast majority of them contain a very young stellar component (between 0.1 and 1 Gyr), i.e. they suffer from a very recent burst of star formation.

Let us now discuss the above results in the context of dynamical models of galaxy interaction, having in mind that the formation of shell structures of long duration is a sort of constraint hinted by the observational data.

Among the N-body and/or SPH dynamical models dealing with the formation of shells, merger models are in general unable to produce shells of long duration. Dupraz & Combes (1986) may produce shells that last longer than in other merger models (Quinn 1984; Hernquist & Quinn 1989), depending on the initial conditions. However these models fail in predicting the radial distribution of shells in the best studied shell galaxies. Only the inclusion of the dynamical friction (Dupraz & Combes 1987), that slows down the merging process between the two galaxies, may produce long lasting shells with the correct distribution. Nevertheless, these models predict also the presence of a double nucleus in the galaxy hosting the shells. Noteworthy, in the present sample of shell galaxies only ESO 240-100, out of twenty objects, is characterized by the presence of a double nucleus.

Furthermore, SPH simulations of mergers by Weil & Hernquist (1993) show that during the shell structure phase, a large fraction of the gas content (if present) falls soon onto the nucleus whereby star formation is likely to occur. In these models shells are as old as the last central star forming event. Similar predictions are found in the models by Kojima and Noguchi (1997), in which the time delay between the end of the star-burst and the shell formation is very short ( Gyr). This prediction hardly matches the occurrence of shell structures and relatively old star-bursts indicated by line strength diagnostic.

Alternatively, dynamical models producing shells via weak interaction event (Thomson & Wright 1990, Thomson 1991) predict much longer lifetimes for the shells up to Gyr. This type of model requires, however, the presence of a thick disc structure in the host early-type galaxy. Furthermore, shells do not develop if the galaxies experience multiple interactions.

Rampazzo et al. (1998b) analyzed the isophotal structure of pair galaxies in our sample and found that a large fraction of them show a disc-like structure. Therefore, if the interaction generating shells is the weak interaction mechanism of Thomson & Wright (1990) and Thomson (1991), we would expect to see them associated to pair galaxies where the disc-like structure is often present. Among the pair galaxies of our sample, only a few possess shell structures, namely RR 225a, RR 225b and RR 278a to which the system NGC 1549+NGC 1553 (the difference among the two systemic velocities is 31 km s^-1: Longhetti et al. 1998b) in the Dorado group can be added.

Finally, let us comment on the observational hint that despite the large uncertainty affecting the indices we have derived, many pair galaxies seem to be characterized by very young bursts.

Among others, RR278a is particularly remarkable as it has the H+K(CaII) index as high as 1.95 and most likely H lines in emission, suggesting ongoing star formation. Another interesting case is NGC 1553 in the center of which Trinchieri et al. (1997), using narrow band imaging, have found H emission. Longhetti et al. (1998a,b) did not detect H emission probably because it was too faint or it was filling the absorption feature. If so, also the value of the H+K(CaII) line may be contaminated by H in emission. It follows from all this that burst in NGC 1553 could be as recent as in pair galaxies with shells.

What the implications are as far as the star formation history in pair galaxies is concerned?

Prior to any other consideration, one has to clarify the question whether the pair galaxies in question are bound or unbound systems. The criteria used to select pairs of galaxies are designed to isolate objects whose interaction potential energy is by far larger than that with any other nearby galaxy. Nevertheless, there remains the controversial problem of the relative percentage of bound vs. unbound pairs (see e.g. Junqueira & de Freitas Pacheco 1994 and reference therein).

In recent years, efforts have been made to disentangle the problem of the physical reality of pairs by means of techniques different from the classical statistical methods. For instance, the X-ray diffuse background has been proposed as a probe of the gravitational potential well and physical reality of a pair/group in turn.

The NGC 2300 group, dominated by the pair NGC 2300/ NGC 2276 (Mulchaey et al. 1993, 1996) and surroundend by X-ray diffuse background could be taken as an example of a bound system. In contrast, the absence of an X-ray diffuse background in K 416 pair of galaxies (Rampazzo et al. 1998a), the analog of NGC2300 as far as morphology is concerned, suggests this system is an unbound encounter.

In this context, the fact that the majority of pair members in our sample underwent a recent star formation event (in some of them even going on now) suggests that either large reservoirs of gas must be present to maintain active star formation (if these galaxies are bound systems on periodic orbits) or most of the pair galaxies in the sample are experiencing unbound encounters and undergo an occasional (perhaps unique) burst of stellar activity.

© European Southern Observatory (ESO) 1999

Online publication: April 19, 1999
helpdesk.link@springer.de