Astron. Astrophys. 352, L51-L56 (1999)

2. Source selection and previous observations

The method we are using to find distant radio galaxies is based on the empirical correlation between redshift and observed spectral index in samples of low-frequency selected radio sources (e.g., Carilli et al. 1999). Selecting radio sources with ultra steep spectra (USS) dramatically increases the probability of pinpointing high-z radio galaxies, as compared to observing radio galaxies with more common radio spectra. This method, which can to a large extent be explained as a K-correction induced by a curvature of the radio spectra, has been shown to be extremely efficient (e.g., Chambers, Miley & van Breugel 1990; van Breugel et al. 1999a).

We constructed such a USS sample (; ; De Breuck et al. 1999b), consisting of 669 objects, using several radio catalogs which, in the southern hemisphere, include the Texas 365 MHz catalog (Douglas et al. 1996) and the NVSS 1.4 GHz catalog (Condon et al. 1998).

As part of our search-program we observed TN J1338-1942 () with the ESO 3.6m telescope in 1997 March and April (De Breuck et al. 1999a). The radio source was first identified by taking a 10 minute band image. Followup spectroscopy then showed the radio galaxy to be at a redshift of , based on a strong detection of Ly, and weak confirming C   1549 and He   1640. At this redshift its derived rest-frame low frequency (178 MHz) radio luminosity is comparable to that of the most luminous 3CR sources.

More detailed radio information was obtained with the VLA at 4.71 GHz and 8.46 GHz on 1998 March 24, as part of a survey to measure rotation measures in HzRGs (Pentericci et al. 1999). We detect two radio components ( mJy; mJy) separated by in the field of the radio galaxy (Fig. 1). The bright NW component has a very faint radio companion ( mJy) at to the SE. Our present observations show that all components have very steep radio spectra with , , and . The proximity and alignment of such rare USS components strongly suggests that they are related and part of one source. While further observations over a wider frequency range would be useful to confirm this, for now we conclude that TN J1338-1942 is a very asymmetric radio source, and identify component C at and with the radio core. Such asymmetric radio sources are not uncommon (e.g., McCarthy, van Breugel & Kapahi 1991), and are usually thought to be due to strong interaction of one of its radio lobes with very dense gas or a neighboring galaxy (see for example Feinstein et al. 1999).

Fig. 1. 4.85 GHz VLA radio contours overlaid on a Keck band image. The cross indicates the position of the likely radio core at 8.5 GHz, which appears offset from the galaxy by () along the radio axis. Contour levels are -0.23, -0.17, -0.12, 0.12, 0.15, 0.17, 0.20, 0.35, 1.45, 5.8, and 29 mJy/beam

We also obtained a band image with the Near Infrared Camera (NIRC; Mathews & Soifer 1994) at the Keck I telescope on UT 1998 April 18. The integration time was 64 minutes in photometric conditions with seeing. Observing procedures, calibration and data reduction techniques were similar to those described in van Breugel et al. (1998). Using a circular aperture of 3", encompassing the entire object, we measure (we do not expect a significant contribution from emission lines at the redshift of the galaxy). In a 64 kpc metric aperture, the magnitude is , which puts TN J1338-1942 at the bright end, but within the scatter, of the relationship (van Breugel et al. 1998).

We determined the astrometric positions in our band image using the USNO PMM catalog (Monet et al. 1998). We next used the positions of nine stars on the band image in common with the Keck band to solve the astrometry on the band image. The error in the relative near-IR/radio astrometry is dominated by the absolute uncertainty of the optical reference frame, which is (90% confidence limit; Deutsch 1999). In Fig. 1, we show the overlay of the radio and band (rest-frame band) images. The NW hotspot coincides within of the peak of the band emission, while some faint diffuse extensions can be seen towards the radio core and beyond the lobe. The positional difference between the peak of the band emission and the radio core is (), which suggests that the AGN and peaks of the band and Ly emission may not be co-centered.

© European Southern Observatory (ESO) 1999

Online publication: November 23, 1999
helpdesk.link@springer.de