The data discussed here and those for Mrk 477 (Heckman et al 1997) clearly demonstrate that hot, massive stars (O and B) are present in the nuclei of these Seyferts. These starbursts are very compact. The effective (half-light) radii are of-order 102 parcsecs, from about 55 pc in IC 3639 to 200 pc in NGC 5135. They are therefore about an order-of-magnitude more compact than the circumnuclear starburst in NGC 1068 (Bruhweiler et al 1991) and the circumnuclear rings reported by Colina et al (1997) in other type 2 Seyfert galaxies, which are several kpc in diameter.
The UV absorption features due to stellar winds (SiIV, CIV, and NV) suggest that the duration of the starburst was short. Is it possible that the starburst formed in an instantaneous event of star formation? Assuming a typical velocity dispersion for these galaxies of 200 km s-1 and a diameter of the nuclear starburst of 330 pc, 160 pc, 405 pc and 110 pc for Mrk 477, NGC 7130, NGC 5135 and IC 3639, respectively, the crossing times within the starburst are 1.6 Myr, 0.8 Myr, 2.0 Myr and 0.5 Myr. Thus, arguments of causality show that it is dynamically possible for these to be effectively instantaneous bursts of star-formation (i.e. the duration of the star-formation could be significantly less than the evolutionary lifetime of a single generation of massive stars). However, it is remarkable how close to the limit our derived durations for the starburst are; the star formation that took place must have been extremely efficient.
Our comparison with evolutionary synthesis models indicates that we are seeing a starburst in a particular evolutionary state. The age estimated is between 3 and 5 Myr for NGC 7130, NGC 5135 and IC 3639, and 6 Myr for Mrk 477. The equivalent widths of the Balmer recombination lines are also compatible with a burst of 5 Myr. This age range represents a special phase of the star formation event, which is when the most massive stars evolve to the Wolf Rayet (WR) phase. For solar metallicity and Salpeter IMF, it is expected that a population of WR stars should be present in the starburst. In Mrk 477, a possible WR signature was detected in the optical spectrum. The feature, emission underlying broad HeII 1#14686, has a luminosity compatible with the presence of 14#14WR stars. We detect HeII in the other three Seyfert nuclei, but the width indicates that this is likely a nebular line. Nebular HeII has been detected in high-excitation HII galaxies (Campbell, Terlevich, & Melnick 1986; e.g. IZw18 Izotov et al 1997) and giant HII regions (e.g. NGC 2363 González-Delgado et al 1994) and appears to be excited by WR stars (Schaerer & Vacca, 1997). However, the ratio HeII/H10#10 (70#700.02) in these star-forming objects is much lower than that observed in our three Seyfert 2 nuclei (0.14, 0.16 and 0.13 for NGC 7130, NGC 5135 and IC 3639, respectively). This suggests that the HeII nebular emission (like the [NeV] emission) is excited by the hard radiation produced by the hidden type 1 Seyfert nucleus. We do not detect any other WR signatures in the optical, such as NV 1#14604,4620 or NIII 1#14634,4640. Another WR star signature in the UV region is the HeII 1#11640 stellar wind line. Unfortunately, our GHRS spectra do not include this wavelength. However, Thuan (1984) detects this line in the IUE spectrum of NGC 7130 and NGC 5135, and he attributes it to WR stars.
Could the WR signatures in the optical region be diluted by a previous star formation event that is now in the post-starburst phase? If the contribution of this population to the optical spectrum were larger than to the UV, the optical signatures could be diluted significantly, but not the UV ones. The detection of the CaII triplet could support this argument. Evolutionary synthesis models predict that CaII is very strong in absorption if the star formation event is in the post-starburst phase with an age between 10 and 20 Myr (García-Vargas et al 1997), due to the evolution of the massive stars to the red supergiant phase. However, for ages less than 7-8 Myr the number of red supergiants in the starburst is too small to produce significant CaII absorption. Thus, the strength of the CaII triplet in NGC 7130 would indicate a burst of 10-20 Myr. However, for these ages, the stellar wind absorption lines in the UV could not be formed, because these lines are formed only in the stellar winds of O stars. To reconcile these two features, we require continuous star-formation, or a significant contribution to CaII by bulge stars, or a burst with a finite time duration or a model of two bursts. In this last case, a young burst (3-5 Myr) to explain the UV features and an older one (10-20 Myr) to explain the CaII triplet is required. A similar solution was found to explain the WR features and the CaII triplet detected in the super stellar cluster A of NGC 1569 (González Delgado et al 1997). On the other hand, the UV morphology of these starbursts (see Figure 4b, 6b and 7b) suggests that the knots detected could be super star clusters. They could be formed in multiple bursts with different ages. Probably, the UV continuum and the stellar absorption features are dominated by the youngest cluster, and the near-infrared continuum and the CaII triplet absorption lines by the oldest one. The spread in age of the clusters is determined by the time duration of the burst. Thus, the integrated light of the starburst could be dominated by O massive stars at the UV wavelengths and by cooler stars at the red and near-infrared.