The HQS provides objective
prism plates for 567 fields of the northern
high galactic latitude sky with 
and direct plates for most of them. In many cases
two prism plates per field are available. The field size of the plates
is 
and the spectra have a non-linear dispersion
with 1390 Å/mm at H
. The Kodak IIIa-J emulsion is used, giving
a wavelength coverage of 
between the atmospheric UV-limit
at 
and the emulsion sensitivity cut-off at 5400 Å.
The limiting
magnitude for the spectral plates is 
but can differ between
the worst and the best plates up to 1 magnitude. The upper brightness limit
due to saturation is reached between 12 < B < 14 for the objective
prism plates. The direct plates have
a limit of 
.
Figure 1: Digitized objective prism spectrum
of 1RXS J095703.2+563138. The left image shows a part of the
digitized objective prism plate (low resolution scan) around the X-ray
position. The right image displays the spectrum
from the high-resolution and low-resolution (inset) scan
The plates are digitized with a PDS microdensitometer. To fasten the digitization process the prism plates are scanned with low resolution, providing density spectra consisting typically of 15-20 pixels. Interesting spectra can be rescanned individually with a tenfold higher resolution. The left part of Fig. 1 (click here) shows a cut of the digitized spectral plate around the position of an arbitrary RASS source, and the right part presents the digitized spectrum in high and low resolution.
The digitized spectral plates are calibrated in the Johnson B-band with
photometric sequences from the Guide Star Photometric Catalogue
(Lasker
et al. 1988) and faint sequences from various sources. Down to
the
photometric accuracy varies in the range 0.2-0.5 mag depending on the
sequences available. Below the quality of the calibration
degrades gradually. The calibration of extended optical sources
(galaxies) yields brightnesses systematically too faint often more
than 1mag.
The HRC marks such brightnesses with an "e'' for extended after the magnitude.
The digitized direct plates provide coordinates with
an accuracy of 
and images with moderate to good resolution
(2'' - 4'').
The identification process is performed in several steps. Originally the
positions of X-ray sources were provided by the first processing of the
survey data (RASSI, Voges et al. 1992a).
This system collected all RASS sources in great
circle strips 
wide passing through the ecliptic poles. Data
from neighbouring strips were combined to create a RASS
X-ray source file for each HQS field.
Correlations with stellar positions
from the SIMBAD catalogue revealed that the RASSI positions have a
1
-error of 
(Voges 1992). For the search of
counterparts we adopted an error circle with radius 
or
, whichever is larger and where is the
standard deviation of the X-ray position as given by the first RASS processing.
In order to account for the larger positional error of the spectra in
dispersion direction the error circle was enlarged by 
(2.3'') in this direction (forming an ellipse).
The selection of optical counterparts is made field by field. The first step is the registration of all optical sources inside the X-ray error circle. The method of registration was changed in 1995 to extract the full information from the spectral plates down to their limit. Prior to 1995 all spectra inside the error circle were automatically selected from the digitized low-resolution spectral database and rescanned with higher resolution. Then, objects from the direct plates, too faint to be visible on the spectral plates, were added. Since 1995 the automatic selection was made on the direct plates and scans with high resolution were made at the transformed positions. This procedure has the advantage to provide additional density spectra with low signal to noise for objects which failed to enter the digitized low resolution spectral database.
The classification of the optical sources is made basically with the high resolution objective prism spectra. These spectra are displayed on a video screen and are classified interactively. Only galaxies and galaxy clusters are mainly recognized due to their appearance on the direct plates. The classification criteria are described in Sect. 3 (click here) and a detailed description of the catalogue is given in Sect. 4 (click here).
The RASS-BSC is based on a second processing of the RASS data,
yielding 
80000 sources (RASS-II).
The main differences in the second data processing as compared to the first
are as follows. The photons were not collected in strips but rather merged in
1378 sky-fields of size 
, taking full advantage of
the increasing exposure towards the ecliptic poles. Neighbouring fields
overlap by at least 
, in order to ensure detection of sources at
the field boundaries, which was a problem in the first processing.
The candidate list for the maximum-likelihood analysis was enlarged by
lowering the threshold values for the two preceding source detection
algorithms, and the acceptance criteria were changed to allow very soft and
very hard sources to be included. These might have been rejected previously if
the likelihood in the broad energy band was too low. The calculation of the
spline-fitted background map was improved, resulting in more accurate
count rates. Finally a new aspect solution was incorporated; this reduced
drastically the number of sources with erroneous positions and morphology.
In order to ensure the quality of the source detection process a visual
inspection of all RASS-BSC
sources was performed (Voges et al. 1997a). While the total number
of sources increased from 50000 (RASS-I) to 80000 (RASS-II) there are
only a few sources in the RASS-BSC, which were not already detected as RASS-I
sources.
Typical error radii for RASS-II sources
are 
. Due to the relatively large error circles
(
) adopted for the identification of RASS-I sources, the
error circles of the corresponding RASS-II sources usually fall within the
RASS-I error circles. We then only updated our catalogue with the new
X-ray positions. Checks were made, if the criteria leading
to a particular classification were altered by the new X-ray data. Finally,
a few sources were discarded from the final catalogue because the error
circles of the two RASS processings do not overlap.