2 Observations and reductions

The Fabry-Perot (FP) observations of HCG 31 were carried out on 1 November 1997 at the f/7.5 Cassegrain focus of the 2.1 m telescope of the Observatorio Astronómico Nacional in San Pedro Mártir, B.C., México using the UNAM Scanning Fabry-Perot Interferometer PUMA (Rosado et al. 1995). The scanning FP interferometer is based upon an ET-50 Queensgate Instruments etalon with a servo-stabilization system. The H$\alpha$ line is observed in interference order 330, within a free spectral range of 19.89 Å, and sampled at 48 steps of 0.43 Å separation (18.9 km s^-1). The spectral resolution of these observations is 38.4 km s^-1. The effective finesse obtained with this setup was 24. Since HCG 31 subtends only 2$^\prime$ and we were careful to place it at the center of the instrument's field of view (10$^\prime$) when acquiring the data cubes, the effective finesse of the observations is the same. In any case, PUMA does not show important variations of the effective finesse across the field. A 30 Å interference filter, centered at the wavelength of redshifted H$\alpha$, was used to isolate the H$\alpha$ emission line from HCG 31. The detector was a 1024 $\times$ 1024, thinned Tektronix CCD. The image scale was 0 $.\!\!^{\prime\prime}$59 pixel^-1, yielding a 10$^\prime$ field of view. In order to increase the S/Nof the observations, the detector was used with a $2\times 2$ pixel binning.

We obtained two data cubes of HCG 31 in H$\alpha$ with exposure times of 48 min each (1 min/channel). These data cubes were co-added to enhance the S/N of the faint regions. The seeing was about 1 $.\!\!^{\prime\prime}$2. The transparency conditions were rather good and we obtained both cubes during dark time. Nevertheless, we corrected both data cubes for transparency variations before co-adding them. This was done using two field stars (located outside the region shown in Fig. 1) and we verified that the profiles were similar.

We obtained wavelength calibration data cubes of a Ne lamp at the beginning and end of the observations, which also serve to check for possible equipment flexures. Since the redshifted H$\alpha$emission of HCG 31 differs from the wavelength of the calibration lamp, a phase shift correction was applied, which amounted to a shift in the zero-point of the velocities of 22 km s^-1. The CIGALE software package (Le Coarer et al. 1993) was used to apply this phase shift correction, to remove cosmic rays, calibrate in wavelength, and construct the radial velocity cubes. We also used some routines from the Image Reduction and Analysis Facility (IRAF) for parts of the data reduction.

The multi-object spectroscopy of the galaxies in HCG 31 was obtained with the 2.1 m telescope of the Observatorio Astrofísico Guillermo Haro in Cananea, México on 10 January 1999. The LFOSC spectrograph was used, which is a transmission spectrograph employing a grism as the dispersing element (Zickgraf et al. 1997). The spectrograph's dispersion was approximately 5.5 Å/pix and the spectra typically spanned from 4000 Å to 6600 Å, though the exact spectral range varied according to the object's position within the spectrograph's field of view. The detector was a $576\times 384$ EEV CCD. Since the spectrograph has very poor response in the blue, no order-sorting filter was used. The objects were selected for spectroscopy using focal plane masks made from previously-acquired images. The only restriction for object selection is that they may not be aligned in declination since the dispersion axis is oriented east-west. Pairs of holes were cut for object and sky at the same right ascension so as to ensure identical spectral coverage. The holes were 3 $^{\prime\prime}$ in diameter.

Figure 2: The velocity map of HCG 31 in H$\alpha$. The velocities are heliocentric velocities. Panel a) shows the velocities for galaxies A, C, E, and F; panel b) shows the velocities for galaxy B; and panel c) shows the velocities for galaxy G. In all cases, the colour bars show the total range of radial velocities. The ellipses are again meant to illustrate the positions of galaxies A, C, E, F, and the tidal candidates, as in Fig. 1. In general, both our radial velocities and velocity profiles are in good agreement with those of Rubin et al. (1990). The continuity of the kinematics for galaxies A, C, and E is particularly striking, and probably indicates that these three components are a single entity. Galaxy Q was not detected in our Fabry-Perot data cubes, while the H$\alpha$ line from galaxy D is shifted outside our filter by this galaxy's much higher radial velocity (Hickson et al. 1992). In all panels, north is up, east is to the left, and $1\hbox {$^{\prime \prime }$ }= 281$ pc.

Figure 3: The velocity dispersion map of galaxies A and C in H$\alpha$. The colour scale shows the range in velocity dispersion values. Here, we characterize the velocity dispersion as the standard deviation of the Gaussian fit to the velocity profile. Only galaxies A and C showed significant internal variations in velocity dispersion, so they are the only ones shown here. The contours show the total H$\alpha$ flux from the Fabry-Perot observations. As usual, north is up, east is to the left, and  $1\hbox {$^{\prime \prime }$ }= 281$ pc.

The objects selected for spectroscopy are shown in Fig. 1, with the exception of galaxy Q (Rubin et al. 1990), which is to the north of the field. The total integration time was 1.5 hours.

The standard stars HD 19445, HD 74721, and HD 109995 were observed for flux calibration. The standard stars were observed through masks cut for various object fields.

The multi-object spectroscopy was reduced using the IRAF software package, specifically the noao.imred.specred package. The bias images were first combined and the result subtracted from all of the images. Pixel-to-pixel variations were then removed using spectra of the internal lamp. Subsequently, the sky was subtracted from each object. Spectra of the Ne-Ar lamp were used for wavelength calibration. Finally, the flux calibration was made using the standard star observations.