Galaxies are complex systems where gas, dust, stars and dark matter deeply interact with each other for a very long time. The emerging spectrum is therefore sensitive to a number of parameters as mass, age, metallicity, star formation history, dust, geometry etc. Despite this complexity, many groups have tried to model the galaxy emission, and the observed optical spectra can now be accurately reproduced (e.g. Bruzual & Charlot, 1993; Worthey, 1994; Fioc & Rocca-Volmerange, 1997; Barbaro & Poggianti, 1997; Tantalo et al., 1996). This is increasingly important as more high-redshift galaxies are discovered and important quantities as age, mass and star formation history are estimated by comparing the integrated properties of the galaxies with the predictions of spectrophotometric models.
These models are calibrated on local galaxies and the first test on their properties is the comparison of the modeled spectra with the observed ones over a large wavelength range. Until now, this comparison could only be done in the optical and near-UV region where representative galaxy spectra are available (e.g., Pence, 1976; Coleman, Wu & Weedman, 1980; Kennicutt, 1992a, 1992b; Kinney et al., 1996, hereafter K96), while photometric points are used for all the other wavelengths. The aim of this work is to extend the range where a full comparison is possible towards longer wavelengths, between 1 and 2.4 µm. This is in particular important because large differences are present between the different models especially in this wavelength range where the relative contribution of dwarf, giant and supergiant stars is poorly known (e.g., Charlot, Worthey & Bressan 1996). These stellar populations are characterized both by different depths of the absorption lines and different shapes of the near-IR continuum (see, for example, Pickles, 1998; Lançon et al., 1999) and therefore by studying breaks, colours and lines of a full spectrum it is possible, in principle, to separate their contribution.
Near-IR spectra can also be used to compute the k-corrections of an unevolving galaxy population, and, therefore, to study the luminosity functions and the differential evolution of the galaxies (Pence, 1976; Coleman et al. 1980, Frei & Gunn 1994, Poggianti, 1997). The near-IR k-corrections are becoming more and more important as new IR-optimized telescopes are becoming available, as the Very Large Telescope, the Large Binocular Telescope and the Next Generation Space Telescope. Complete spectra between UV and near-IR wavelength can also play a role in the measure of the redshifts of large samples of galaxies, both from photometric data (for a review see Yee 1998) and even from spectroscopic data (eg., Glazebrook, Offer and Deeley, 1998).
In this paper we show the results of the observations of 28 galaxies of morphological type between E and Sc, while later classes will be the subject of a future work. Target selection, observations and data reduction are described in sec. 2. Great care was taken in matching the near-IR spectra to the optical ones by using the observed UV-optical-IR colours, as described in sec. 3 and sec. 4. In sec. 5 we derive the k-corrections in the J, H and K band. In the following sections, we compare the total spectra to the predictions of some spectrophotometric models for elliptical galaxies and use the observed absorption lines to study the dominant stellar populations. Spectra and k-corrections are available in electronic form.