2. Sample selection and data analysis
We have selected a sample of bright radio-quiet objects from the Veron-Cetty & Veron (1993 ) and the Hewitt & Burbidge (1993 ) quasar catalogues by requiring the radio (flux measured at 5 GHz) to optical (V band) broad-band spectral index to be flatter than 0.3 . For objects where no radio flux measurement was available, an upper limit of 25 mJy was assumed. Note that, through this selection criterium, objects weaker than magnitude in the v band are only included in the sample when radio flux measurements are available and the radio flux is below 25 mJy, essentially resulting in a cutoff of the sample at . All objects show broad emission lines and have absolute optical magnitudes brighter than and are thus classified as quasars.
These objects were cross-correlated both with the list of ROSAT PSPC pointings from the ROSAT archive and with the IUE ULDA database of low resolution spectra. Objects were included in the sample when (1) at the time of the analysis (summer 1995), UV spectra were available in the ULDA database and the ROSAT data had entered the public domain, (2) the objects were located within 44 minutes of arc from the optical axis of the ROSAT observation, (3) they were not hidden by the PSPC support structure, (4) at least 180 PSPC source counts had been measured. The final sample defined in this way which was investigated in our further analysis consists of 31 objects, listed in Table 1.
Table 1. The AGN Sample ROR: ROSAT sequence number (PSPC). T is the exposure time in ks, z is the redshift of the object, and R the radio flux in mJy. The UV flux of IUE (LWP and SWP camera) is given in Jy. CR is the ROSAT PSPC count rate [cts/s] in the energy range 0.1 to 2.4 keV. See Sect. 2 for details on the broad-band spectral indices .
Depending on the distance of the object from the optical axis of the ROSAT PSPC detector, X-ray counts were extracted within a circular area centered on the source position with radii ranging from 1 to 5 minutes of arc. For each object, the background was accumulated from an annulus centered on the source position from which areas contaminated by other sources had been excluded. The data were binned into background subtracted count rate spectra with, depending on the number of source counts, 6 to 23 spectral bins in the energy range from 0.1 to 2.4 keV (channel 11 to 240) and a dead time correction and a correction for telescope vignetting was applied. The binning was performed such that a S/N of at least 5 was achieved in each bin.
The broad-band spectral indices and used for the sample selection and in our statistical analysis were calculated from the source frame luminosity densities at 5 GHz, 2500 Å, and 2 keV, respectively, which were determined following Zamorani et al. (1981 ) and Marshall et al. (1983 ), and applying corrections for reddening following Seaton (1979 ). The resulting values are listed in Table 1.
Data from the IUE ULDA database were used to determine continuum flux values in each of the frequency bands 130-140, 140-155, 166-170, and 170-185 nm (SWP camera) and 245-260, 260-275, 275-290, and 290-305 nm (LWP/LWR camera). The continuum flux level was determined by removing emission and absorption lines using a sliding window technique with window widths of 0.4, 0.8, 1.6, 3.2, 6.4, and 12.8 nm. De-reddening of the resulting continuum fluxes was performed by using (Bohlin et al. 1987 ) with galactic values as derived from 21 cm radio measurements (Stark et al. 1992 and Elvis et al. 1989 ).
© European Southern Observatory (ESO) 1997
Online publication: April 8, 1998