1. Introduction

Since the availability of the IRAS database, several surveys have been devoted to the search of ultraluminous infrared galaxies (ULIRG), a class of objects with an infrared luminosity above 10¹² , comparable to the bolometric luminosity of optically selected quasars. Some ULIRG samples have directly been selected from IRAS flux catalogs, using distance-independent criteria, such as optical to infrared flux ratios (Clements et al. 1996b) or infrared colors (Lawrence et al. 1986; Kim 1995), making them good ULIRG candidates. Confirmation of their high luminosity was then obtained from spectroscopic follow-ups. Others have been extracted from complete, flux limited, IRAS galaxies redshift surveys. The two more important ones so far published are the IRAS Bright Galaxy Sample (BGS, Soifer et al. 1986) for galaxies with a declination above and a 60 m\ flux lower limit of 5.24 Jy, extended to the whole sky in the Extended Bright Galaxy Sample (Sanders et al. 1996), and the deeper 1.9 Jy redshift survey by Strauss et al. (1992), which covers the entire sky. In the Bright Galaxy Sample, Sanders et al. (1988a) have found 10 ULIRGs, with an infrared luminosity greater than . These galaxies have afterwards been extensively studied in various wavelength ranges (e.g. Carico et al. 1988; Majewski et al. 1993). We present here a sample of ULIRGs, which was extracted from the list of Strauss et al. (1992), restricted to the southern hemisphere, and complete for 60 m fluxes greater than 3 Jy. Northern ULIRGs of the 1.9 Jy survey have recently been presented by Murphy et al. (1996). The 24 Southern Ultraluminous Infrared Galaxies (hereafter referred as SULIRG) in the sample have been studied by means of optical and infrared imaging and spectroscopy.

Despite the high number of papers devoted to Ultraluminous Infrared Galaxies (see the recent reviews by Sanders & Mirabel 1996 and Moorwood 1996), many questions about their nature remain unanswered. In particular the power source of their enhanced infrared luminosity - starburst, AGN, combination of both - is unknown. Their morphological class is even matter of debate. Although it is widely admitted that most ULIRGs are interacting or merging systems, the actual proportion varies greatly according to the the authors (Sanders et al. 1988a; Melnick & Mirabel 1990; Leech et al. 1994). To that respect, the choice of the ULIRG parent catalogue is of prime importance, since it may introduce biases in statistical studies. Extracting an ULIRG sample from a list of faint IRAS sources, distributed over a limited portion of the sky, Leech et al. (1994) have selected distant objects for which a morphological classification is dubious. Using infrared warm objects, Sanders et al. (1988b) have preferentially chosen active galaxies. To that respect the BGS and 1.9 Jy redshift surveys appear to be more likely to provide with less biased samples of ULIRGs.

Section 2 describes our sample, and Sect. 3, the observations and the data reduction. Some general results dealing with the morphology and the spectral classification of the SULIRGs are presented in Sect. 4. Notes on individual objects are given in Sect. 5.