3 The data

3.1 Observations  

The sample galaxies were observed in the course of several observing campaigns, from December, 1992 until February, 1997. Most of the data were collected at the TIRGO telescope (Zermatt, CH) in 1992, 1993, 1994 and 1997; some galaxies were observed at the NOT telscope (La Palma, Canary Islands) in 1995 and some at the VATT telescope (Mt. Graham, Az) in 1996. The instrument used at all telescopes was the Arcetri NIR camera ARNICA, equipped with a 256$\times$256 pixel NICMOS3 detector (Lisi et al. 1993, 1996; Hunt et al. 1996). The plate scale was 0.97, 0.55, and 0.45 arcsec pixel^-1 at the TIRGO, NOT and VATT respectively.

All the galaxies were observed by alternating frames on the source and on adjacent empty sky positions, integrating $\sim$$1^{\rm m}$ in each position. Typical total integration times on source were $\sim$$5^{\rm m}$ in all bands, yielding ($1\sigma$) surface magnitude limits of 20.8, 21.3, and 20.6 H-mag arcsec^-2 at the TIRGO, NOT and VATT respectively (Hunt & Mannucci 1998). With a typical sky brightness of 14 H-mag arcsec^-2, these limits correspond to $0.1 \, - \, 0.2$% of the "background''.

3.2 Data reduction

All image reduction was performed with IRAF and the STSDAS packages.

The flat fields for each source exposure were obtained by averaging the sky frames acquired immediately before and after it, upon removal from each sky frame of eventual stars in the field. The use of medians of larger numbers of sky frames as flat fields ("superflats''), despite their lower noise, turns out to be effective only in case of exceptional stability of the atmospheric emission. Due to the high sky brightness, the requirements on flatness for mapping the outer regions of galaxies are quite strict. On average the resulting flatness (low-spatial-frequency noise only) of our final images is about 0.01% ($1\sigma$), negligible relative to the high-spatial-frequency noise which amounts to 0.1 - 0.2% as noted in Sect. 3.1. Such image quality allows a substantial improvement in signal-to-noise of the elliptically-averaged profiles (see Sect. 3.4).

The - typically four - flat-fielded source frames were subsequently cleaned for bad pixels, registered, and rescaled by an additive term to a common median level. They were then combined to obtain the median frame, after clipping deviant values in each position on the detector. The final step was the subtraction of the background, which was estimated on the source frame itself upon automatic exclusion of star and galaxy pixels.

3.3 Photometric calibration  

A star from the list of northern standards by Hunt et al. (1998) was observed once about every hour in the three bands. In each band five frames were obtained with the star in different positions on the detector. A flat field for each standard star position was obtained using the "clipped'' median of the remaining four positions. Standard-star frames were also cleaned for bad pixels, but not averaged together. For every standard star observation, aperture photometry was performed in each position within a circle centered on the star, after subtracting the sky background. The radius of the aperture was typically 4 times the seeing FWHM; the background value was evaluated as the median in a circular annulus between radii 5 and 8 FWHM. The instrumental magnitudes were (to first order) corrected for atmospheric extinction using the coefficients reported in Table 1 (from Hunt & Mannucci 1998); units are mag airmass^-1. An average zero point was then computed for each night of observations in every band and used to calibrate the relative galaxies.

  

Table 1: Extinction coefficients

  

Table 2: Summary of observations

All the data presented here, magnitudes, surface magnitudes, and diameters, take into account only correction for atmospheric extinction; no correction has been applied for Galactic and internal extinction nor for the effects of redshift (K correction, and angular size - z relation). A summary log of the obervations is presented in Table 2, where for each observing night: Column 1 indicates the telescope used (N for NOT, T for TIRGO, V for VATT); Column 2 is the date of the following day; Column 3 is the nightly averaged zero point in the H band; Column 4 is the $1\sigma$ uncertainty on the calibration as derived by the set of standard-star observations for the night; Column 5 is the average FWHM of the seeing measured on the standard-star images; Column 6 is the list of sample galaxies observed that night.

H-band multiaperture photometry of 11 of the sample galaxies is available from Bothun et al. (1985), who used a single-element, InSb photometer; the galaxies are UGC 646, 673, 732, 820, 919, 927, 1013, 1033, 1094, 12486, 12359. Similar data are also available from Baffa et al. (1993) for UGC 57, 60, 98, 975, 1094, 1100, 1302, 1411, 2548, 12527, 12666. In order to compare these with our results, we have performed multiaperture photometry on our images using their aperture values, for a total of 24 measurements to compare with Bothun et al. and 55 with Baffa et al. The mean difference between the results by Bothun et al. and ours is $+0.04 \pm 0.08$ mag (standard deviation of the set, not of the mean), a quite comfortable result; the comparison with Baffa et al. yields a difference of $+0.07 \pm 0.14$ mag. In neither case is a systematic difference clearly discernible.

3.4 Brightness profiles  

For each of the final galaxy images the coordinates of the center were determined by fitting a Gaussian to the center. Brightness profiles were then extracted by fitting elliptical contours of increasing galactic radius, keeping the center position fixed, with surface brightness $\mu_{\rm H}$, ellipticity $\epsilon=1-b/a$, and position angle PA as free parameters. The profiles were sampled for increasing values of the major semiaxis a, from 1 to 10 pixels in steps of 1 pixel, and then outward with a $10\%$ geometric progression $a_{i+1}=1.1 \times a_i$.Regions of the image containing close companions or foreground stars were edited out from the fitted area.

We also provide an estimate of the seeing FWHM for each final image of a galaxy. This was determined in most cases using stars in the field of the galaxy image. When this was not possible, stars were selected in the images immediately before or after the one considered. The resulting profiles for $\mu_{\rm H}$, $\epsilon$, and PA are reported to the right of the corresponding image in Fig. 4.