4 Radial velocity measurements

The radial velocities cz have been determined by two different, independent methods: first, "automatic'' identification of the absorption and/or emission lines and fourier cross correlation with template spectra using the rvsao package; second, "by eye'' identification of absorption lines and computation of cz with the task rvidlines. Both methods were applied to all measured spectra.

A number of our spectra show OII, H$\beta$, and OIII in emission. Emission lines were removed from the spectra and a cross-correlation analysis was carried out. Radial velocities were also independently measured for these emission lines.

For the first method high signal-to-noise spectra of selected galaxies (NGC 1407, NGC 1426, and NGC 1700) were employed as templates in the the cross-correlation analysis. These templates were observed with the same instrument. A synthetic spectrum, as in Way et al. (1998), was also used. All templates were rebinned to the same resolution as the fiber spectra and were corrected for heliocentric velocity. The task xcsao crosscorrelates the emission-line-cleaned object spectra with the template spectra and generates a reliability factor called the R value which depends on the amplitude of the cross correlation peak. A low R value (R<4) indicates that one should control the results by eye. The "best'' velocity, as judged by the highest R value was adopted. In cases where strong emission lines, especially the OII line, were present we calculated in addition their velocities with the task (emsao). In some cases where the absorption lines were very weak, but the emission lines clearly visible, the estimated emission line velocity could be used to improve the absorption line detection with xcsao. The absorption line velocities $v_{\rm cross}$ as well as the emission line velocities $v_{\rm em}$ are presented in Table 2.

The second method is based on the following steps: the Balmer jump and the close K and/or H lines were identified by eye and marked for the IRAF task rvidlines which compares the marked lines with a list of absorption lines. All identified lines with a sufficient signal-to-noise were used for the velocity calculation. The centering algorithm used for the position of the lines calculates the wavelength where the total flux of the absorption feature is divided into equal halfs. The resulting velocities $v_{\rm rvid}$ are presented in Table 2.

  

Figure 2: This plot shows the differences of the two independent methods of velocity determinations. The rms error of the scatter of the most reliable identifications (Q = 1 and 2) is about 150 km s-1

The results of both methods were compared. In 18 of the 115 spectra the signal-to-noise was too low to determine a velocity by either of the methods. In 68 spectra the velocities of both methods agree within $\pm$500 km s^-1. Figure 2 shows the differences of the two determinations. The standard deviation is about 130 km s^-1, comparable to the rms errors of the fit to individual velocity determinations. One of our galaxies, the nucleated dwarf elliptical NGC 1396, was measured in both fiber sets. The signal-to-noise is one of the highest in our sample. The velocities of the "automatic'' identification agree very well: 865 $\pm$ 17 and 871 $\pm$ 32 km s^-1. The "by eye'' identification yields 870 $\pm$ 44 and 918 $\pm$ 52 km s^-1. Within the errors all values are consistent and agree with the published value by Jones & Jones (1980): 894 $\pm$ 120 km s^-1.

In 9 cases where the object is located very close to a Fornax elliptical or S0 galaxy, most probably the light, and thus the velocity of the giant was measured instead that of the object. In 5 of these spectra a second velocity was found indeed. In Table 1 we give the name of the Fornax galaxy, the ratio of the Fornax giant galaxy to target object light in a 2 arcsec aperture, the velocity of both components, and the identification quality as defined in the next section.

  

Table 1: Objects whose spectra have a dominant contribution by light of a close Fornax giant galaxy

  

Table 2: Radial velocities of galaxies in and behind the Fornax cluster

 

Table 2: continued