The 100 kpc scale tidal features imaged in H I emanating from starbursting systems suggest that the interaction and starburst timescales are quite different. For example, the starburst in the IR luminous merger Arp 299 has an age of < 30 Myr while the 180 kpc tail was launched about 700 Myr ago (Hibbard & Yun 1999). Similarly, the tails of the Antennae (Fig. 1) suggest an interaction timescale of ~ 500 Myr. This object has a population of star-clusters with an age of ~ 500 Myr as well as a population that is currently forming massive stars (Whitmore & Schweizer 1995). These observations suggest that interaction induced starbursts are not isolated to either first periapse (when the tails are launched) or the final merger, but rather are episodic (cf Noguchi 1991). While the closeness of the nuclei of the ultraluminous IR mergers suggests that the most intense starbursts occur when the progenitor nuclei are coalescing, it does not necessarily follow that the bulk of the stars are formed during this short-lived phase.
This fact has important repercussions for the expected observational characteristics of merger remnants. If most of the interaction-induced starformation takes place at the moment of final coalescence, the burst population is expected to be confined to the inner few 100 pc of the remnant, leaving an anomalously bright central core (Mihos & Hernquist 1994) with the characteristics of a younger, more metal enriched population. The lack of such signatures in shell galaxies is taken to mean that they could not have formed via major mergers (Silva & Bothun 1998). However, if much of the post-interaction population formed over the whole ~ 1 Gyr timescale of the merger, then the "burst" population will be spread more widely through the remnant, leaving much more subtle observational signatures.