The most prominent emission lines are L9#9 and CIV 1#11549. Other high-ionization lines such as NIV] 1#11486, SiIV+OIV] 1#11397+1403 and NV 1#11240, and low-ionization lines, such as CII 1#11335 are also present at least in some of the nuclei. These lines are prominent in IC 3639 and almost absent in NGC 5135. This makes the spectrum of IC 3639 similar to that of Mrk 477 (Heckman et al 1997). However, the absorption features are stronger in IC 3639 than in Mrk 477. The emission lines are mainly produced in the Narrow Line Region (NLR) of the AGN, but some contribution to the CIV, SiIV and NV comes from the starburst component (stellar wind), and this is probably the case in IC 3639. However, the starburst is the dominant component in NGC 5135, where CIV and SiIV are weak and NIV] is absent. NGC 7130 represents an intermediate case, since NIV] is absent but CIV and SiIV are stronger than the contribution expected from the starburst (see next section). L9#9 is strong in the three galaxies. This line can be produced by the AGN or by the starburst component.
In IC 3639, all the emission lines CIV, NIV], NV, and L9#9 show a double peak. The observed separation between the two peaks of CIV and NV lines are equal to the expected separation between the two components of CIV and NV doublets, and the peaks are at the rest wavelength of these lines. On the other hand, the two peaks in L9#9 fall 170 km s-1 redshifted and 620 km s-1 blueshifted with respect to the systemic velocity. The fluxes are 49#49 and 50#50erg s-1 cm-2, and the FWHM's (not corrected for instrumental resolution) are 2.05 Å and 1.7 Å for the red and blue components, respectively. The profiles look asymmetric, with the blue wing of the redshifted component and the red wing of the blueshifted component absorbed. This double peak could be produced by absorption in outflowing gas driven by the AGN or by the starburst.
In NGC 7130, L9#9 appears asymmetric. The blue side of the profile drops rapidly, and the peak of the line is redshifted 0.5 Å (120 km s-1) with respect to the systemic velocity. Even though the shift is lower than the shift observed in some starburst galaxies (González Delgado et al 1998; Kunth et al, 1997), the shape of the profile suggests that we are seeing here the same phenomenon. The explanation is that the neutral gas responsible for the absorption is blueshifted (outflowing) with respect to the systemic velocity. The flux of the L9#9 emission line is 51#51 erg s-1 cm-2 and the FWHM is 52#52 Å (not corrected for instrumental resolution).
In NGC 5135, L9#9 is broader than in the other two galaxies, with a 53#53 Å and a flux of 54#54 erg s-1 cm-2. The center of the line is absorbed, leaving double emission peaks at -410 km s-1 and 180 km s-1 relative to the systemic velocity. Again, this is suggestive of absorption by outflowing gas.
No broad component (width of several thousand km/s) is detected in L9#9 or CIV, indicating that scattered light from a hidden Seyfert 1 nucleus does not contribute significantly to the UV spectrum of these Seyfert 2 nuclei. To illustrate this we have compared the spectrum of IC 3639 (which is the galaxy with the largest contribution from the Seyfert component, as shown by its emission lines) with the spectrum of the average Seyfert 1 and low redshift QSO observed by IUE. Following Heckman et al (1995), we iteratively decrease the fractional contribution of the Seyfert 1 spectral component to the UV continuum until the result is consistent with the lack of broad emission in the L9#9 and CIV lines. Figure 9 shows that less than 5% of the UV continuum can be due to scattered light from a hidden Seyfert 1 nucleus. Because CIV in emission is weaker in the other two Seyferts than in IC 3639, we expect that the contribution of scattered light from a hidden Seyfert 1 nucleus in NGC 5135 and NGC 7130 is completely negligible in the GHRS aperture.