2. Observations and data reduction
2.1. The spectroscopic observations
The spectroscopic observations of our sample galaxies were carried out at the ESO 1.52-m Spectroscopic Telescope at La Silla on February 15-19, 1994.
The telescope was equipped with the Boller & Chivens Spectrograph. The No. 26 grating with 1200 was used in the first order in combination with a slit. It yielded a wavelength coverage of 1990 Å between about 5200 Å and about 7190 Å with a reciprocal dispersion of 64.80 Å mm-1. The instrumental resolution was derived measuring after calibration the FWHM of 22 individual emission lines distributed all over the spectral range in the 10 central rows of a comparison spectrum. We checked that the measured FWHM's did not depend on wavelength, and we found a FWHM mean value of 2.34 Å. This corresponds to Å (i.e., 51 at 5800 Å and 46 at 6400 Å). The adopted detector was the No. 24 20482048 Ford CCD, which has a m2 pixel size. After an on-chip binning of 3 pixels along the spatial direction, each pixel of the frame corresponds to 0.97 Å .
The long-slit spectra of all the galaxies were taken along their optical major axes. At the beginning of each exposure the galaxy was centered on the slit using the guiding camera. Repeated exposures (typically of 3600 s each) did ensure several hours of effective integration without storing up too many cosmic rays. Some long-slit spectra of 8 late-G or early-K giant stars, obtained with the same instrumental setup, served as templates in measuring the stellar kinematics. Their spectral classes range from G8III to K4III (Hoffleit & Jaschek 1982).
The typical value of the seeing FWHM during the observing nights, measured by the La Silla Differential Image Motion Monitor (DIMM), was . Comparison helium-argon lamp exposures were taken before and after every object exposure. The logs of the spectroscopic observations of galaxies and template stars are reported in Tables 2 and 3, respectively.
Table 3. Log of spectroscopic observations (template stars).
2.1.1. Data reduction
Using standard MIDAS 2 routines, all the spectra were bias subtracted, flat-field corrected by quartz lamp exposures, and cleaned from cosmic rays. Cosmic rays were identified by comparing the counts in each pixel with the local mean and standard deviation, and then corrected by substituting a suitable value.
A small misalignment was present between the CCD and the slit. We measured a difference of pixel between the positions of the center of the stellar continuum near the blue and red edge of the spectra. When measuring the stellar kinematics, the tilt had to be removed. This was done by rotating the spectra by a suitable angle () before the wavelength calibration. We noticed however that the sharp line profile of the emission lines was spoiled by the rotating algorithm. For this reason no rotation was applied when measuring the ionized-gas kinematics.
The wavelength calibration was done using the MIDAS package XLONG. We determined the velocity error possibly introduced by the calibration measuring the `velocity curve' of a sample of 24 OH night-sky emission lines distributed all over the spectral range. The velocity did not show any significant dependence on radius, indicating that the wavelength rebinning had been done properly. We found a mean deviation from the predicted wavelengths (Osterbrock et al. 1996) of 2 .
After calibration, the different spectra obtained for a given galaxy were co-added using their stellar-continuum centers as reference. For each spectrum the center was assumed as the center of the Gaussian fitting the mean radial profile of the stellar continuum. The contribution of the sky was determined from the edges of the resulting galaxy frames and then subtracted.
2.1.2. Measuring the gas kinematics
The ionized-gas velocities () and velocity dispersions () were measured by means of the MIDAS package ALICE. We measured the [N II] lines (6548.03, 6583.41 Å), the H line (6562.82 Å), and the [S II] lines (6716.47, 6730.85 Å), where they were clearly detected. The position, the FWHM, and the uncalibrated flux F of each emission line were determined by interactively fitting one Gaussian to each line plus a polynomial to its local continuum. The center wavelength of the fitting Gaussian was converted into velocity in the optical convention ; then the standard heliocentric correction was applied. The Gaussian FWHM was corrected for the instrumental FWHM, and then converted into the velocity dispersion . In the regions where the intensity of the emission lines was low, we binned adjacent spectral rows in order to improve the signal-to-noise ratio, , of the lines.
We expressed the variation of the r.m.s. velocity error as a function of the relevant line ratio. In order to find the expression for , we selected the same 16 night-sky emission lines in the spectra of NGC 2775, NGC 3281, IC 724 and NGC 4845. Such night-sky emissions were chosen to have different intensities and different wavelengths between 6450 Å and 6680 Å (i.e. the wavelength range of the observed emission lines of the ionized gas) in the four spectra. We derived the sky spectra by averaging several rows along the spatial direction in a galaxy-light free region. Using the above package, we interactively fitted one Gaussian emission plus a polynomial continuum to each selected sky line and its local continuum. We derived the flux F and the FWHM of the sample lines, taking the ratio as the signal S. For each galaxy spectrum the noise N was defined as the r.m.s. of the counts measured in regions of the frame where the contributions of both the galaxy and the sky lines were negligible. The resulting range was large (). For each sample emission line we then measured, by means of an automatic procedure, the night-sky `velocity curve' along the full slit extension. The wavelengths of the emissions were evaluated with Gaussian fits and then converted to velocities. The radial profiles of the sky-line velocities were then fitted by quadratic polynomials. We assumed the r.m.s. of the fit to each `velocity curve' to be the velocity error. Fig. 1 shows the good agreement between the distributions of the measurements taken in the 4 different spectra. In log-log scale, the relation is well represented by a straight line, that corresponds to:
(least-squares fit). This result agrees with Keel's (1996) relation , based on numerical simulations. Once the relevant ratio of the emission had been derived, we obtained for each velocity measurement of the ionized-gas component by means of Eq. 1. The gas velocities derived independently from different emission lines are in mutual agreement within their errors .
The ionized-gas velocities and velocity dispersions from [N II] (6548.03, 6583.41 Å), [S II] (6716.47, 6730.85 Å), and H are reported in: Tables 4-7 for NGC 2179; Tables 10-14 for NGC 2775; Tables 17-21 for NGC 3281; Tables 24-26 for IC 724; Tables 29-33 for NGC 4698; and Tables 36-40 for NGC 4845. Each table reports the galactocentric distance r in arcsec (Col. 1), the observed heliocentric velocity v and its error in (Col. 2), the velocity dispersion in (Col. 3), the number n of spectrum rows binned along the spatial direction (Col. 4), and the signal-to-noise ratio of the emission line (Col. 5). The H, [N II] and [S II] kinematics of the sample objects are plotted in Figs. 2-4.
The final ionized-gas kinematics is obtained by averaging, at each radius, the gas velocities and velocity dispersions derived independently from the different emission lines. The gas velocity () and velocity error () are respectively the -weighted mean velocity and its uncertainty. The gas velocity dispersion () and velocity-dispersion error () are the mean velocity dispersion and its uncertainty. (No error is given when only one velocity dispersion measurement is available.)
The kinematics of the ionized gas is reported in: Table 8 for NGC 2179; Table 15 for NGC 2775; Table 22 for NGC 3281; Table 27 for IC 724; Table 34 for NGC 4698; and Table 41 for NGC 4845. Each table reports the galactocentric distance r in arcsec (Col. 1), the mean heliocentric velocity and its error in (Col. 2), the mean velocity dispersion and its error in (Col. 3). The ionized-gas velocity and velocity-dispersion profiles are plotted in Fig. 5 for all our galaxies.
2.1.3. Measuring the stellar kinematics
The stellar velocities () and velocity dispersions () of the sample galaxies were measured from the absorption lines in the wavelength range between about 5200 Å and 6200 Å. We used an interactive version of the Fourier Quotient Method (Sargent et al. 1977) as applied by Bertola et al. (1984). The K0III star HR 5100 was taken as template: it has a radial velocity of -0.9 (Wilson 1953) and a rotational velocity of 10 (Bernacca & Perinotto 1970).
The spectra of the galaxies and the template star were rebinned to a logarithmic wavelength scale, continuum subtracted, and masked at their edges by means of a cosine bell function of length. At each radius the galaxy spectrum was assumed to be the convolution of the template spectrum with a Gaussian broadening function characterized by the parameters , and . They respectively represent the line strength of the galaxy spectrum relative to the template's, and the line-of-sight stellar velocity and velocity dispersion. The parameters of the broadening function, and consequently the stellar kinematics, were obtained by a least-squares fitting in the Fourier space of the broadened template spectrum to the galaxy spectrum in the wavenumber range . In this way we rejected the low-frequency trends (corresponding to ) due to the residuals of continuum subtraction and the high-frequency noise (corresponding to ) due to the instrumental resolution. (The wavenumber range is important in particular in the Fourier fitting of lines with non-Gaussian profiles, see van der Marel & Franx 1993 and Cinzano & van der Marel 1994). In deriving the above kinematical properties, the regions Å and Å were masked because of contamination from bad subtraction of the night-sky emission lines of [O I] ( Å) and Na I ( Å).
The measured stellar kinematics is reported in: Table 9 for NGC 2179; Table 16 for NGC 2775; Table 23 for NGC 3281; Table 28 for IC 724; Table 35 for NGC 4698; and Table 42 for NGC 4845. Each table reports the galactocentric distance r in arcsec (Col. 1), the heliocentric velocity and its error in (Col. 2), the velocity dispersion and its error in (Col. 3). The stellar velocity and velocity-dispersion profiles are plotted in Fig. 5.
2.2. The photometric observations
The observation in the Cousin band of NGC 2179 was performed on March 11, 1997 at the 1.83-m Vatican Advanced Technology Telescope (VATT) at Mt. Graham International Observatory. A back-illuminated 20482048 Loral CCD with 15m2 pixels was used as detector at the aplanatic Gregorian focus, f/9. It yielded a field of view of with an image scale of pixel-1 after a pixel binning. The gain and the readout noise were 1.4 e- ADU-1 and 6.5 e- respectively. We obtained 3 images of 120 s with the R 3.48-inch square filter.
The data reduction was carried out using standard IRAF 3 routines. The images were bias subtracted and then flat-field corrected. They were shifted and aligned to an accuracy of a few hundredths of a pixel using field stars as reference. After checking that the point spread functions (PSFs) in the images were comparable, they were averaged to obtain a single R image. The cosmic rays were identified and removed during the averaging routine. A Gaussian fit to the intensity profile of field stars in the resulting image allowed us to estimate a seeing PSF FWHM of .
The sky subtraction and the elliptical fitting to the galaxy isophotes were performed by means of the Astronomical Images Analysis Package (AIAP: Fasano 1990). The sky level was determined by a polynomial fit to surface brightness in frame regions not contaminated by galaxy light; then it was subtracted out from the total signal. The isophote fitting was performed masking the frame's bad columns and the bright field stars. We then obtained surface brightness, ellipticity, major-axis position angle, and the Fourier coefficient of the isophote's deviations from elliptical as a function of radius along the major axis. No photometric standards were observed. Thus the absolute calibration was made using the photometric quantities edited by Lauberts & Valentijn (1989) in the same band. We set the surface brightness of an isophote with semi-major axis of to the value R-.
© European Southern Observatory (ESO) 1999
Online publication: February 23, 1999