2 Observations and reduction

2.1 Observations

Ferguson (1989) lists 340 likely Fornax cluster members within an area of $\sim$40 sq deg centered on the cluster. Of these, 14 are classified as elliptical galaxies brighter than $B_{\rm T} = 15.0$ mag. We were able to obtain kinematical data for 12 of these galaxies making our investigation 86% complete at that limiting magnitude. In addition, we have obtained kinematical data for FCC 119, (classified as a S0 galaxy by Ferguson, and for which we give only the central velocity dispersion (CVD) and $V_{\rm max}$).

The spectra were obtained over two runs in November and December 1996 using the blue arm of the Double Beam Spectrograph attached to the Australian National University's 2.3 m telescope at Siding Spring Observatory. Dichroic #3 was used, providing greater than 95% transmission at all wavelengths between 4000-6000 Å. A 1200 l mm^-1 grating was used, with a dispersion of 0.555 Å$\:$ pixel^-1 over a range of 1000 Å$\:$ centered on 5200 Å. The CCD used was a SITe chip (1752$\times$532) with 15 $\mu$m pixels. The spatial scale on the chip was 0.91$\hbox{$^{\prime\prime}$}$ pixel^-1. We used a spectrograph slit of 2$\hbox{$^{\prime\prime}$}$ on the sky and of length greater than the spatial extent of the CCD. The best collimator focus gave a FWHM for the arc lines of 2.7 pixels or 1.50 Å, giving a resolution of 86 km s^-1 at 5200 Å. During the observations the seeing varied from 1.5 to 2.5 arcsec.

The observed galaxies are listed in Table 1, along with their photometric parameters (taken from Caon et al. 1994, we also refer to this paper for a detailed photometric description). We obtained our estimates of the galaxy major-axis PAs based on careful inspection of the position angle (PA) profiles (Caon et al. 1994). Only our estimate for the major-axis PA of NGC 1419 differed from the value adopted in Caon et al. (1994), where we have adopted 50$^{\circ}$rather than 65$^{\circ}$.The number of spectra for each galaxy and the total exposure times used are also listed in the table. The spectra were exposed in 30-45 min blocks. A Ne-Ar lamp was observed before and after each exposure for wavelength calibration and four template stars (of spectral types from G8III to K3III) were observed at the beginning, middle, and end of each night. The usual dome and sky flats were taken, as were bias frames and measurements of the dark current.

  

Table 1: Galaxy sample and photometric parameters

The columns respectively list the R.A. and Dec. of each galaxy, the NGC number (where available) and Fornax Cluster Catalogue number (FCC; Ferguson 1989), morphological type according to Ferguson (1989), total apparent blue magnitude, effective half-light radius, and associated blue surface brightness at that radius and major-axis position angle (measured E from N). The number of spectra and the total exposure time for each object are also given. The photometric parameters are taken from Caon et al. (1994).

2.2 Data reduction

The data reduction included, besides the usual CCD cosmetics, noise removal along the spatial direction by adaptive filtering techniques (Richter et al. 1992). After wavelength calibration and sky subtraction, the spectra were normalized to the continuum, obtained by fitting with a 6th order polynomial. The reliability of the sky subtraction was iteratively tested by comparing the brightness profiles of the sky-subtracted spectra to the profiles from Caon et al. (1994). The slit function, as derived from twilight spectra, varied by at most 0.5%, so that no further correction was needed. The low-frequency residual variations were reduced by filtering in Fourier space. Finally, the spectra were processed by the Fourier Correlation Quotient (FCQ) technique (Bender 1990). All four template stars were used, and the relative results were compared to test their consistency: as expected, no relevant difference was detected (see Bender 1990). In order to test the reliability of the observed features, we compared the velocity dispersion profiles (VDPs) and the rotation curves (RCs) obtained by adding the different spectra of each individual galaxy before and after the final stage of processing (the FCQ), obtaining fully consistent results. As a further test, we also reduced the data without filtering and, although the data were obviously more noisy in the fainter regions, the observed features were still present. A lower limit of $\sim$35 km s^-1 in the measurable velocity dispersion, due to the instrumental setup, was also verified on the template stars. The uncertainties in the data are those derived by the FCQ procedures, basically arising from the fit of the broadening function, and are explained in Bender (1990).