2 MEFOS observations and data reduction

A total of 104 redshift measurements for 99 galaxies within a circular region of $30\hbox{$^\prime$}$ around the radio position of the cluster cD, $\mbox{RA} = 20^{\rm h} 58^{\rm m} 39\hbox{$.\!\!^{\rm s}$}0$ and $\mbox{Dec} = -28\hbox{$^\circ$}15\hbox{$^\prime$}22\hbox{$^{\prime\prime}$}$ (Tadhunter et al. 1993), were obtained using the MEFOS multifiber spectrograph at the 3.6-m ESO telescope at La Silla (Chile). The observations were carried out on May 23-26, 1995 during an observing run whose main target was the dwarf galaxy population of the Centaurus cluster (see Stein et al. 1997). The MEFOS instrument has a circular field of view of $1\hbox{$^\circ$}$ and 29 fiber arms which carry two spectral fibers of $2\hbox{$.\!\!^{\prime\prime}$}5$ aperture for simultaneous object and sky acquisition, and one image fiber of $36\hbox{$^{\prime\prime}$}\times 36\hbox{$^{\prime\prime}$}$ for the interactive repositioning of the spectral fibers. A grating with 300 lines mm^-1 was used to produce spectra in the range between 3800 and 6100 Å with a typical resolution of ca. 10 Å. The detector was a TI $512\times 512$ CCD chip.

The raw CCD spectra were reduced using the MIDAS package through several steps which include cleaning from defects, cosmic ray removal, flat-fielding, one-dimensional extraction, and wavelength calibration using a He-Ne lamp before and after each exposure. The sky subtraction was performed subtracting from each one of the object spectra the mean of the output of all the fibers positioned on blank sky positions from the corresponding exposure. Prior to sky subtraction the signal of each spectrum (including the sky spectra) was scaled with respect to the intrinsic transmission efficiency of the corresponding fiber, which had been determined using the average over the observed fields of the signal under the OI emission line at 5577.4 Å.

After the final one-dimensional spectra had been extracted, velocities were computed either from emission lines or from absorption lines, or from both. Emission-line redshifts were obtained from galaxies with at least two clearly visible emission lines (mostly OII, H$\beta$, and OIII). Their redshifted positions were determined from fits with a Gaussian superposed onto a quadratic polynomial approximating the local continuum. The final redshift of a galaxy was then computed as the unweighted mean over the n emission lines present in its spectrum. Since the errors in the redshift measurement of each single line are essentially dominated by uncertainties in the wavelength calibration (Stein 1996), individual measurement errors were taken equal to 100 $\mbox{km\,s}^{-1}$, independently of line strength. Accordingly, an uncertainty of $100/\sqrt{n}$ $\mbox{km\,s}^{-1}$ was assigned to emission-line redshifts. Absorption-line redshifts were obtained using the standard cross-correlation algorithm described by Tonry & Davis (1979). This technique requires the previous removal of both galaxy emission lines and strong night-sky lines, and the transformation of the spectral continuum to a constant level of zero. Special care was taken that the continuum subtraction did not create spurious features of low spatial frequency which could be confused with broad, superposed absorption lines. For the determination of the zero-point shift one single template was constructed by merging 20 galaxy spectra with high S/N and well known redshifts. Only normalized cross-correlation peaks of height 0.25 or larger were considered as significant. Both emission-line and cross-correlation redshifts were then corrected to heliocentric values.

The 26 galaxies observed also by Stein (1996) in the field of A3733 with the OPTOPUS spectrograph were used to determine the scaling factor of the internal errors estimated in the cross-correlation procedure, resulting in external errors of typically 40 - 50 $\mbox{km\,s}^{-1}$. These same galaxies gave a mean velocity difference of -23 $\mbox{km\,s}^{-1}$ (the MEFOS redshifts being typically smaller), consistent with zero to within the reported measurement errors. Only for 5 of our galaxies we could measure both emission-line and cross-correlation redshifts. Again, an excellent consistency was found between the two kinds of measurements.

The cross-correlation and emission-line radial velocities for the 99 galaxies observed with the MEFOS spectrograph in the field of A3733 are listed in Cols. (4) and (5) of Table 1, together with their estimated external errors. Columns (6) and (7) give the same information for the 39 galaxies observed with the OPTOPUS instrument by Stein (1996). Column (8) lists the final radial velocities and their estimated uncertainties which result form a weighted average of the data in Cols. (4)-(7). The combination of these two samples gives a total of 112 entries, which will be used in the following section to examine the kinematical properties and structure of A3733. This is about three times the number of galaxies used in the previous study by Stein (1997). The first three columns of Table 1 contain the celestial coordinates for the epoch B1950.0 and the $b_{\rm J}$ magnitudes from the COSMOS catalog kindly provided by H. MacGillivray. The completeness in apparent magnitude of the final dataset is high, with percentages of 100, 92, 75, 63, and 50% of all known (COSMOS) galaxies in the same region of the sky covered by our observations at the $b_{\rm J}$ magnitude limits of 17.0, 17.5, 18.0, 18.5, and 19.0, respectively.