Once the catalogue is created, the sources have to be identified.
First, the X-ray positions are cross-correlated with the
digitised objective prism plate, and all spectra inside the error circles are
evaluated. An error radius 
is used with the
additive term included to allow for systematic errors. 
is the
standard deviation calculated by the source detection algorithm.
Of all candidates thus selected, B-magnitudes are computed from their
objective prism spectra which cover the entire Johnson B band. A method was
developed (Köhler 1991) which uses an internally determined
characteristic curve, calibrated externally with photometric sequences. The
spectral information is used to synthesize the B band. More details are given
in Engels et al. (1994). Well above the plate limit, these are accurate
to a few tenths of magnitude, with the error rapidly increasing towards fainter
objects. Exact (
) positions are obtained by identification of the
candidates on the direct plate. During this step, obvious artifacts among the
spectra are removed, and further optical counterparts below the objective
prism plate limit can be found.
Provided the spectra are brighter than 
, a
classification is possible: early stars can be distinguished from late or very
late ones, galaxies and AGN are usually recognizable.
At 
, only the raw forms of the spectra can be seen;
they are classified as weak and extremely blue (these are usually faint AGN),
blue (again, many of these are AGN, although the fraction of non-AGN
grows) or red. For a certain amount of sources, no classification is possible,
e.g. saturated spectra or overlaps. Examples of the respective objective prism
spectra and a detailed description of this identification step
are given in Bade et al. (1997).
All optical counterparts thus found are cross-checked with published
catalogues (Véron-Cetty & Véron 1996;
SIMBAD
,
NED) and the HQS archive.
22 objects from the original optical quasar sample could be identified
as
X-ray sources. In total, the following results were
obtained:
| 24 | known AGN, |
| 14 | known stars, |
| 2 | clusters of galaxies (one candidate), |
| 51 | AGN candidates (classified as AGN or extremely |
| blue), | |
| 21 | galaxy candidates, |
| 50 | stellar candidates (including saturated spectra), |
| 23 | faint blue objects, |
| 17 | faint red objects, |
| 19 | unidentified spectra (including overlaps), |
| 270 | visible on the HQS direct plate only, |
| 83 | empty fields (i.e. no counterpart on the HQS plate). |
Several error circles contain more than one optical counterpart. For the
determination of the most plausible one, all information is put together.
The ratio of X-ray to optical flux 
is crucial in
cases where both stellar and non-stellar objects are found; the former have
mostly 
while the latter have
. Raw spectral information could be
helpful in classifying white dwarfs (very soft X-ray spectra) or galaxy
clusters (hard X-ray spectra). In those cases where a known AGN falls inside
the error circle, it has been automatically taken as the proper identification.
Some X-ray sources in the field HS 47.5/22 have been previously detected,
either in the RASS (Bade et al. 1995; Voges, private communication) or with
Einstein. For these, the published identification is used.
Follow-up observations were performed during a 5 nights period in March 1994 with the 3.5m telescope at Calar Alto/Spain. Candidates were selected by decreasing X-ray count rate in the attempt to obtain a flux limited quasar sample. X-ray sources with obvious stellar identifications were ignored, and some special sources with lower count rates were included .
The 3.5m telescope was equipped with the focal reducer (f/2.7) and a TEK
CCD (
pix, 
). With this instrumentation it
is possible to switch between direct imaging and spectral mode. Due to
non-photometric weather and moderate seeing (
), a rather
wide slit (
) had to be used. For direct imaging, Johnson BVR
filters were mounted.
Since many candidates were faint (
),
and since our foremost interest was redshift determination, the
low-dispersion grism (
) was used to obtain a sufficient
signal to noise in the shortest possible time. The large wavelength range
thus covered (
) includes normally at least two
emission lines, so that the resulting redshifts are accurate to better than
0.01. Line shape analysis, on the other hand, is practically impossible with
this low resolution (FWHM 
). Acquisition exposures were made
with the B-filter to obtain better magnitudes than those derived from the
objective prism plates.
Spectra of a He lamp have been taken to get an absolute wavelength scale.
In a few cases, the wavelength calibration was corrected with the sky
absorption lines at 
and 
. Spectrophotometric
and photometric standards were observed two or three times per night for flux
and magnitude calibration. However, due to non-photometric weather condition,
no absolute flux calibration could be achieved.
54 X-ray sources could be observed. Five of them had no optical counterparts
on the HQS plates, all of them hard band detections. They might therefore be
distant clusters of galaxies or high redshift quasars, both unlikely to
appear on blue plates. One minute red exposures were taken in search of
counterparts, with a limit of 
. Two exposures showed
an object inside the respective X-ray error circles, of which spectra were
obtained. One could be identified with a 
quasar
(RXJ 0948.5+4751), the other (RXJ 0958.3+4702) with a galaxy at
. Although the missing counterparts and thus extreme
made the remaining three sources even more
interesting, deeper searches had to be omitted in favour of identifying other
objects.
45 X-ray sources are clearly extragalactic. One is the galaxy mentioned above, another is a group of three more or less active galaxies, 41 are AGN. For the remaining two cases, the identification is ambiguous. Both spectra show emission lines on top of a galaxy continuum, but neither resolution nor signal to noise is sufficient to reveal any further details on the object types. These two sources are marked as "ELG'' in Table 6.
Three objects have featureless slit spectra. Their objective prism spectra
resemble those of emission line AGN at similar apparent magnitudes, so a
classification as "BL Lac candidate'' seems plausible. For one object, this
possibility is supported by radio data: RXJ 0950.2+4553 has been detected at
6cm (Gregory et al. 1996), 21cm and 74cm. The other two, however,
appear to be radio quiet. Their ratio of X-ray to optical flux is moderate
compared with Nass et al. (1996), who found that many X-ray selected
BL Lac objects have 
. Although neither
missing radio detections nor moderate (
for both) rules out the BL Lac hypothesis entirely, these facts argue against
it. Therefore both objects are denoted "unidentified'', although it
can be safely claimed by the featureless spectra that they are no ordinary AGN.
The last three objects are stars. The low resolution of the spectra allows no classification beyond the recognition of two M-stars which are easily recognizable by their broad absorption bands.
Table 6 lists the observed objects, their final identification, B-magnitude and
redshift (where available). The AGN spectra are presented in
Fig. 7 (click here), and redshift and magnitude distribution of the AGN are
shown in
Figs. 6 (click here)a and b.
Figure 6: B magnitude a) and redshift b) distribution of the newly identified
AGN. The emission line galaxies and the group (counted as one) are shaded
One additional object was observed by N. Bade with the Russian 6m telescope
near Selenchukskaya, covering the wavelength range 
. This object had originally been selected as AGN candidate
in the HQS but was misidentified during their follow-up observations.
The new spectrum is shown at the end of Fig. 7 (click here) . One emission line
is clearly visible above the noise. Associating it with MgII2798
gives a redshift 
. The absence of other lines supports this
since at that redshift, strong emission lines falls into the observed
spectral range.