2 Spectral observations, data reduction and analysis

2.1 Observations

The spectroscopic data were obtained with the 6m telescope of the Special Astrophysical Observatory of Russian Academy of Science (SAO RAS) during two runs in February and April 1999. The Long-Slit spectrograph (LSS in Table 1) (Afanasiev et al. [1995]) at the telescope prime focus was equipped with a Photometrics CCD-detector PM1024 (with $24\times24~\mu$m pixel size) (PMCCD in Table 1) installed at Schmidt-Cassegrain camera F/1.5. Most of the long-slit spectra ( $1.2\hbox{$^{\prime\prime}$ }\times180\hbox{$^{\prime\prime}$ }$) were obtained with the grating of 325 grooves/mm, giving a dispersion of 4.6 Å/pixel. Additional data were obtained with the grating of 1302 grooves/mm and dispersion 1.2 Å/pixel. For the latter set-up the slit of $2\hbox{$^{\prime\prime}$ }\times180\hbox{$^{\prime\prime}$ }$ was used. The scale along the slit was 0.39 $\hbox{$^{\prime\prime}$ }$/pixel.

The resulting resolution (FWHM) was about 14-15 Å for the first set-up, and about 3.7 Å for the second set-up. Reference spectra of an Ar-Ne-He lamp were recorded before or after each observation to provide wavelength calibration. Spectrophotometric standard stars from Massey et al. ([1988]) were observed for flux calibration at least twice a night.

Observations and data processing in this set-up have been conducted mainly under the software package NICE in MIDAS, described by Kniazev & Shergin ([1995]).

 

 

Table 3: Line intensities in the knot "a'' of VV 432, VV 747 and VV 543W

	VV 432 (a) (4.6 Å/pixel)		VV 432 (a) (1.2 Å/pixel)		VV 747 (a) (4.6 Å/pixel)		VV 543W (4.6 Å/pixel)
$\lambda_{0}$(Å)	$F(\lambda)/F$(H$\beta$)	$I(\lambda)/I$(H$\beta$)	$F(\lambda)/F$(H$\beta$)	$I(\lambda)/I$(H$\beta$)	$F(\lambda)/F$(H$\beta$)	$I(\lambda)/I$(H$\beta$)	$F(\lambda)/F$(H$\beta$)	$I(\lambda)/I$(H$\beta$)
3727 [O II]	2.17 $\pm$ 0.10	2.63 $\pm$ 0.12	--	--	1.08 $\pm$ 0.06	1.57 $\pm$ 0.10	3.79 $\pm$ 0.33	4.31 $\pm$ 0.40
3868 [Ne III]	0.33 $\pm$ 0.03	0.39 $\pm$ 0.04	--	--	0.30 $\pm$ 0.01	0.40 $\pm$ 0.02	--	--
4101 H$\delta$	0.26 $\pm$ 0.02	0.30 $\pm$ 0.04	0.23 $\pm$ 0.03	0.26 $\pm$ 0.04	0.22 $\pm$ 0.01	0.28 $\pm$ 0.02	--	--
4340 H$\gamma$	0.49 $\pm$ 0.02	0.54 $\pm$ 0.03	0.50 $\pm$ 0.03	0.54 $\pm$ 0.04	0.40 $\pm$ 0.01	0.47 $\pm$ 0.02	0.50 $\pm$ 0.07	0.53 $\pm$ 0.10
4363 [O III]	0.11 $\pm$ 0.01	0.12 $\pm$ 0.02	0.09 $\pm$ 0.03	0.09 $\pm$ 0.03	0.09 $\pm$ 0.01	0.11 $\pm$ 0.01	--	--
4861 H$\beta$	1.00 $\pm$ 0.02	1.00 $\pm$ 0.03	1.00 $\pm$ 0.04	1.00 $\pm$ 0.04	1.00 $\pm$ 0.01	1.00 $\pm$ 0.02	1.00 $\pm$ 0.09	1.00 $\pm$ 0.11
4959 [O III]	1.10 $\pm$ 0.02	1.08 $\pm$ 0.02	1.09 $\pm$ 0.04	1.07 $\pm$ 0.04	1.79 $\pm$ 0.18	1.74 $\pm$ 0.18	0.46 $\pm$ 0.08	0.45 $\pm$ 0.08
5007 [O III]	3.36 $\pm$ 0.05	3.28 $\pm$ 0.05	3.25 $\pm$ 0.10	3.18 $\pm$ 0.09	5.49 $\pm$ 0.23	5.27 $\pm$ 0.22	1.26 $\pm$ 0.11	1.23 $\pm$ 0.11
5876 He I	0.09 $\pm$ 0.02	0.08 $\pm$ 0.02	--	--	0.09 $\pm$ 0.01	0.07 $\pm$ 0.01	--	--
6548 [N II]	0.05 $\pm$ 0.01	0.05 $\pm$ 0.01	0.03 $\pm$ 0.01	0.03 $\pm$ 0.01	0.03 $\pm$ 0.01	0.02 $\pm$ 0.00	0.20 $\pm$ 0.03	0.17 $\pm$ 0.02
6563 H$\alpha$	3.39 $\pm$ 0.071	2.74 $\pm$ 0.06	3.39 $\pm$ 0.10	2.77 $\pm$ 0.09	2.34 $\pm$ 0.04	2.79 $\pm$ 0.052	3.35 $\pm$ 0.23	2.88 $\pm$ 0.22
6584 [N II]	0.14 $\pm$ 0.04	0.11 $\pm$ 0.03	0.11 $\pm$ 0.01	0.09 $\pm$ 0.01	0.09 $\pm$ 0.03	0.06 $\pm$ 0.02	0.74 $\pm$ 0.08	0.64 $\pm$ 0.07
6717 [S II]	0.26 $\pm$ 0.02	0.21 $\pm$ 0.01	0.29 $\pm$ 0.02	0.23 $\pm$ 0.01	0.13 $\pm$ 0.01	0.09 $\pm$ 0.01	0.99 $\pm$ 0.11	0.84 $\pm$ 0.10
6731 [S II]	0.17 $\pm$ 0.01	0.13 $\pm$ 0.01	0.20 $\pm$ 0.02	0.16 $\pm$ 0.01	0.10 $\pm$ 0.01	0.07 $\pm$ 0.01	0.60 $\pm$ 0.10	0.51 $\pm$ 0.09
C(H$\beta$) dex	0.275 $\pm$ 0.026		0.265 $\pm$ 0.037		0.445 $\pm$ 0.016		0.190 $\pm$ 0.088
EW(abs) Å	0.100 $\pm$ 1.807		0.050 $\pm$ 2.067		0.080 $\pm$ 0.985		0.100 $\pm$ 0.827
F(H$\beta$)	123.10 $\pm$ 1.81		91.08 $\pm$ 2.54		175.09 $\pm$ 1.60		7.78 $\pm$ 0.50
EW(H$\beta$) Å	120.96 $\pm$ 1.78		128.37 $\pm$ 3.58		125.46 $\pm$ 1.14		13.52 $\pm$ 0.87

¹ Due to calibration problems the intensity of H$\alpha$ is likely a bit overestimated. To check its effect we redone calculation of O/H
for C(H$\beta$) = 0. It increases 12+log(O/H) by 0.01. ² -- the intensity of H$\alpha$ is recalculated from the recombination ratio to H$\beta$.

   
2.2 Data reduction and abundances determination

The data reduction was performed in SAO RAS, using various packages of MIDAS (see Kniazev et al. [2000], for details).

In Table 2 we summarize the main observational parameters of the discussed three VV "nests''. They include the names of the objects, their coordinates for the epoch J2000, the apparent blue magnitudes and the corresponding references, the radial heliocentric velocities, measured in this work with their rms uncertainties, maximal angular sizes, absolute blue magnitudes and the oxygen abundances (12+log(O/H)).

Direct images of studied galaxies, extracted from the DSS and the position of long slit, indicated by bar are presented in Figs. 1a, 2a, 3a. Corresponding 2-D spectra are shown in Figs. 1c, 2c, 3c. The brightness profiles of H$\alpha$ line along the slit, and corresponding velocity curves are illustrated in Figs. 1b, 2b, 3b. In Figs. 1d, 2d, and 3d we present 1-D spectra, extracted from 2-D spectra, which were used for the measurements of line intensities, determination of physical conditions and abundances of H II-regions.

The resulting observed emission line intensities $F(\lambda)$ of various ions relative to H$\beta$, both uncorrected and corrected for interstellar extinction and underlying stellar absorption $I(\lambda)$ (following the procedure described by Izotov et al. [1997]) for the brightest parts of the galaxies are presented in Tables 3 along with the extinction coefficient C(H$\beta$), the equivalent width of absorption Balmer hydrogen lines EW(abs), the equivalent width of H$\beta$ line EW(H$\beta$) and the observed H$\beta$ flux.

For the abundances determination we used the scheme, described in detail by Izotov et al. ([1994], [1997]). The electron temperatures and densities in H II-regions of the observed VV-galaxies and their abundances of O, N and Ne are summarized in Table 4.

For VV 543 the [O III]-line 4363 Å is not detected. Therefore to estimate its metallicity we employ the empirical method (see e.g. Pagel et al. [1979]; McGaugh [1991]; Olofsson [1997]). Its uncertainty for 12+log(O/H) can be as large as 0.2-0.3 dex. We also applied this empirical method to all four individual knots of VV 432.

The ratio [O III]/[N II] of extinction corrected intensities of the lines $\lambda$4959, 5007 Å and $\lambda$6548, 6584 Å enable one to avoid well known ambiguity of the empirical method (Alloin et al. [1979]). Taking into account these ratios, one can use R₂₃ and Q parameters to get more reliable estimate of O/H through the model curves by Olofsson ([1997]). By this way it was obtained for VV 543W the abundance 12+log(O/H) = 8.5 $\pm$ 0.1, which is about 0.3 lower than the value, derived from McGaugh ([1991]) tracks, and seems to be in a better agreement with the abundance, expected for underluminous H II-galaxies.

Figure 1: From top-left to bottom: a) DSS image of galaxy VV 432 with the position of long slit superimposed. The spot "a'' is NE bright region at the edge of the galaxy; b) Brightness profile of H$\alpha$ in relative intensities along the slit and the velocity curve. Only independent along the slit points (accounting for seeing) from the spectra with dispersion 1.2 Å/pixel and 4.6 Å/pixel are shown by arrows and filled boxes respectively. The error bars correspond to $\pm 1\sigma$. c) 2-D spectrum of VV 432 with dispersion 4.6 Å/pixel; d) 1-D spectrum of the brightest NE spot ("a'')

2.3 Velocity curves

To obtain line-of-sight velocity distribution along the slit we use the MIDAS programs, kindly presented to authors by D. Makarov.

To increase the accuracy of derived rotation curves the programs were modified to include additional corrections using close lines in the reference spectrum.

The procedure includes the following steps: 1) a linearisation of 2-D spectrum of the object and of the reference spectrum; 2) a measurement of the position of the H$\alpha$ emission for each of the 250 position rows using Gaussian fitting; 3) an estimation of the background and S/N ratio for the H$\alpha$ emission in each row; 4) a similar measurement of the nearby line Ne I $\lambda$6598.95 Å in the reference spectrum and compiling the table of differences between the laboratory wavelength of this line and the measured one for each row; 5) a polynomial fitting of these differences and a determination of the residual scattering (rms); 6) the application of this fitting polynomial to correct the measured wavelengths of H$\alpha$. The resulting error of wavelengths measurements is determined by quadratic summing of the fitting rms, found in the previous step, and the measurement error of velocity for each point along the slit, estimated from Gaussian fitting. The latter varies between 2.5 km s^-1 for the points with the highest signal and 9 km s^-1 for the points with low S/N ratio for dispersion 1.2 Å/pixel; 7) rebinning of H$\alpha$ data along the slit, corresponding to the seeing during the observations (4 or 5 pixels in April and February, respectively).

Below we use only those velocity estimates which satisfy the criteria S/N > 3 and $\sigma_V$ < 15 km s^-1. Resulting high accuracy of corrected observed wavelength enables one to study irregularities of the velocity curve with an amplitude as low as 10 km s^-1 on the angular scale up to 20 $^{\prime\prime}$.

For VV 432, the rotation curves which were obtained with the low and high resolution spectra are presented in Fig. 1b. Their similarity shows that our low dispersion spectra can be used to derive preliminary dynamical parameters of the studied galaxies.