3.2. The rest-frame optical spectroscopy
We have also obtained near-IR spectroscopy for some Lyman-break galaxies in our sample, which allowed us to study the optical rest-frame spectrum of these systems (Pettini et al. 1998, in prep.). These observations have been carried out at the UKIRT 3.8-m telescope with the CGS4 spectrograph and had, therefore, to be limited to the brightest galaxies of our sample. Even in this case, we could essentially only study the nebular emission lines of [OII], H and [OIII] of the galaxies and only barely detect their continua (see Figure 5). Nevertheless, these data have provided very useful complementary information on the amount of extinction and the kinematics of Lyman-break galaxies.
Figure 5. Two examples of near-IR spectra (K band) of Lyman-break galaxies obtained with the UKIRT telescope and the CSG4 spectrograph (Pettini et al. 1998, in preparation). The observed nebular emission lines are labeled. The vertical dashed lines mark the position of major night sky lines. In each case, the continuum emission is just barely detected. The emission lines are resolved and their velocity widths correspond to velocity dispersions in the range of 50 160 km s-1.
There is a broad agreement between the star-formation rates derived from the H fluxes and the ones obtained from the far-UV continuum luminosity density, once moderate corrections for dust obscuration are included. A typical net correction to the observed star-formation rates would be again in the range of 3x to 7x, depending on the adopted obscuration law. This seems to exclude, at least in the galaxies for which near-IR spectroscopy is available, large amounts of dust obscuration, as have been suggested by some (e.g., Meurer et al. 1997). Interestingly, using the Calzetti's law (1997) to compute H fluxes from the UV continuum results in values that are in line with the observed ones only if continuous star formation is assumed, while significantly higher fluxes are found in the case of a burst (Pettini et al. 1998, in preparation).
As discussed above, the near-IR spectra have also shown differences of up to ~ 400 km s-1 between the redshifts of the UV absorption lines and those of the optical nebular emission lines, with the latter having higher redshift than the former and being more likely to be closer to the systemic velocities of the galaxies than the UV absorption lines. The lines are all resolved and their widths, if interpreted in terms of velocity dispersion, yield values in the range 50 160 km s-1. If the kinematics of these lines is a good tracer of the gravitational motions within the galaxies, then combining the velocity dispersion with the sizes derived from the HST images gives dynamical masses in the range 1010-1011 M, comparable to the mass of the Milky Way bulge (Dwek et al. 1995) and to the mass of the innermost few kpc of an L* elliptical galaxy. While the extent to which the optical nebular lines can be used as dynamical indicators is not known, we note that the total masses involved are very likely to be substantially greater, given that the present IR observations sample only the innermost cores of the galaxies, where the star formation rates are highest.