2 Observations and data reduction

Long slit spectra have been obtained during three different runs at the 1.5 m ESO telescope equipped with a Boller & Chivens spectrograph and a 2048² (15 $\mu$m pixel) CCD camera.

The characteristics of the 21 shell galaxies, 30 pair members and 5 "template'' galaxies belonging to the Gonzalez (1993: G93) sample are described in Tables 1, 2 and 3 of Paper I respectively. Observational conditions and spectral parameters for each of the runs are detailed in Tables 4 and 5 of Paper I. Table 5 of this latter reports: the object identification (1), the observing run (2), the slit position angle oriented NE (3), the exposure time (4), the spectrum quality (5) and the portion of the galaxy enclosed in 5$^{\prime\prime}$ in terms of equivalent diameter $D^{\rm B}_{\rm e}$ i.e. the diameter enclosing half of the total light in the B-band taken from Lauberts & Valentijn (1989: ESO-LV).

Galaxy observations were split into consecutive exposures. We typically obtained three successive frames with the same position angle. Calibration frames (He-Ar lamp) were taken before and after each galaxy exposure to allow accurate wavelength calibrations. Each of the galaxy frames has been treated separately. Pre-reduction, cosmic-ray hits removal, wavelength calibration and sky subtraction have been performed using the IRAF package. Multiple spectra were then co-added before being analyzed with a Fourier-fitting code. The Fourier-Fitting technique (see Simien & Prugniel 1997) is similar to that of Franx et al. (1989) for the simultaneous determination of the velocity dispersion $\sigma(r)$, and the rotational velocity V(r) at a distance r from the centre. A parameter $\gamma(r)$, related to the absorption-line contrast with respect to an approximation of the continuum (represented by a third-order polynomial fitting to the spectrum), is also determined. In order to enhance the S/N ratio, adjacent lines were averaged in the outer parts of the galaxies. For a spectrum at radius r, a weight was assigned to the neighboring lines ($r\pm\delta r$) entering the average: for this weight, a gaussian fall-off as a function of $\delta r$ was chosen, with a FWHM value between 1 and 3 pixels (0$.\!\!^{\prime\prime}$6 and 1$.\!\!^{\prime\prime}$8).

Basic data of the stellar kinematics, i.e. heliocentric recession velocity $V_{\rm 0,hel}$, central velocity dispersion $\sigma_0$, and value of $\sigma_{5\hbox{$^{\prime\prime}$}}$ adopted for the correction of line strength indices in Paper I, are collected in Table 1. The value of $\sigma_{5\hbox{$^{\prime\prime}$}}$ has been obtained coadding spectra along the line perpendicular to the spectral dispersion, in the region between -2$.\!\!^{\prime\prime}$5 and 2$.\!\!^{\prime\prime}$5 with respect to the galaxy centre. Table 2 summarizes the measured difference between the systemic velocity of pair members, $\Delta V$, and the projected separation.

In Table 1, detected emission lines are also reported, and the measured EW's are listed in Table 4. EW's have been measured with the same procedure adopted for the atomic indices (see Paper I), using the spectral bandpasses listed in Table 3. Emissions were detected by direct inspection of the spectra, and in the uncertain cases only after coadding. A careful check of the lines positions was done after having redshifted their wavelength to the restframe. Profiles of the galaxies emission lines were fitted with a gaussian curve whose peak was taken as a measure of the lines centres and used to obtain the kinematics of the galaxies gas components. The approximation of a line with a gaussian profile is acceptable for all the measured cases with the only remarkable exception of IC 5063 (RR331a), where a second line component is visible in its profile at the centre of the galaxy (see next paragraphs). We estimate our precision in higher S/N rows better than 10 km s^-1, while we rejected data with errors larger than 30 km s^-1 (0.5 px). Only for ESO 2890150, ESO 2440120/ESO 2440121 (RR 24a/b), ESO 5450400 (RR 62a), ESO 3860040 (RR 278a) and IC 5063 (RR 331a), emissions are extended outside the galaxy centre and it has been then possible to measure a gaseous velocity profile. We obtained the gas velocity dispersion, corrected for the instrumental dispersions, only for the latter objects.

When characterized by "regular'' shapes, the galaxies velocity profiles have been parametrized with a sigmoid along their major axis, in order to obtain the maximum value of the velocity rotation, $V_{\rm max}$. Actually, the majority of the objects shows not regular curves and their $V_{\rm max}$ value has been obtained simply fitting outer data with a line in both sides of the velocity profile. We are aware both of the uncertainty of the latter method and of the fact that the velocity curves reach only the nuclear parts of the galaxies. Table 5 summarizes the informations concerning the rotational support parameters.