In this section we describe some first results, focussing mainly on
areas I, IVac, and Va.
These areas are of particular interest because they represent the extreme
values of 
and of the RASS integration time in our sample.
Several identifications, in particular of stellar sources at low X-ray count
rates, require further confirmation. These observations are still in progress.
The full set of identifications will be published in Paper III.
The results based on our preliminary identifications for these three areas are summarized in Table 5 (click here). For 323 of the 335 sources of the subsample identifications exist, although some of them are still tentative or uncertain. These uncertain identifications will, however, not affect the following conclusions.
With 116 identifications the
stellar counterparts represent a fraction of 
% of the objects.
This is significantly higher than the fraction of
stellar identifications in the EMSS by Stocke et al. (1991)
who found only 26%
stellar sources, but less than the 65%
stellar counterparts in the EXOSAT HGLS Giommi et al. 1991).
Note that 37 Ke/Me stars are among the
stellar sources we identified. Surprisingly, also 3 bright A-type stars were
found very close to the positions of X-ray sources in area I.
In these cases the X-ray emission is attributed to unseen companions.
Two of them exhibit hard X-ray spectra.
Three white dwarfs, including one previously known (WD046-01), were found.
Their X-ray spectra are extremely soft with hardness ratios HR1
. Furthermore three cataclysmic variables were identified.
Table 5: Identifications for areas I, IVac, and Va.
SIM/NED: Identifications based exclusively on the cross-correlation with the
SIMBAD and NED data bases, respectively. "stars'' includes normal stars,
Ke, Me stars, CVs, binaries, and white dwarfs
The percentage of the individual object classes
on the total identifications is clearly varying between the different areas.
The NGP field IVac shows e.g. a much larger fraction of AGN and a smaller
fraction of stars than area I. Since most stellar sources are at relatively
small distances, this effect cannot be explained by the different lines
of sight through the galactic stellar disk alone. This suggests that the
different neutral hydrogen column densities 
play an important
role. Areas I and IVac represent the extreme cases of 
with the
largest column density in area I and the smallest in area IVac
(see Table 2 (click here)). With 80%
the extragalactic sources are the dominant constituent in area IVac which is
similar to the results of Hasinger et al. (1993) for the
Lockman Hole.
In area I only about 25% of the source are of extragalactic nature. With
63% the stellar sources are the dominant part in this area, in
contrast to 12% in area IVac. Area V is intermediate, both with respect to
the ratio of stellar to extragalactic sources and to 
.
With 
the average hardness ratio of AGN in area I is harder than that
in area IVac which is 
. Area Va is intermediate with
. This effect is clearly a
consequence of the different 
values. Likewise, the median
redshifts of AGN, 
, are different: 0.15 in area I and 0.32 in
area IVac.
Although in area Va with 
the median redshift
is not significantly different from area IVac,
we found AGN at higher redshift than in area IVac. This is probably a
consequence of the greater sensitivity due to the longer integration times
in area Va compensating the higher absorption and allowing the detection of
more distant and hence X-ray fainter objects.
A small fraction of sources were preliminarily classified as "multiple''
(see Table 5 (click here)). This identification does not necessarily mean that
more than
one object is responsible for the X-ray emission. It rather denotes source
positions for which our presently available spectra do not allow to
determine a single obvious counterpart for the X-ray source.
The presence of two possible stellar counterparts represents the largest
fraction, 
.
Nearly all of these sources are located in area I which has the
highest contribution of stellar counterparts. In five cases we found two
AGN close to the X-ray position, in one of these even a third
AGN exists which could also contribute to the X-ray emission. We found
several cases in which a known BL Lac object is located in (or projected on)
a cluster of galaxies, as e.g. RXJ0416.8+0105 = H0414+009 in area I.
These objects do not appear in Table 5 (click here) as "multiple'' object
but were classified as BL Lacs.
In the following we will make use of the identifications in the three areas
to derive an estimate for the error of the X-ray positions.
As mentioned in Sect. 3 (click here), we observed candidates in a circle
with 50
to 60
radius which is approximately the
3
error circle calculated by the SASS.
The positions of the optical counterparts were taken from the APM
catalog in most cases. Positions of stars brighter than 6.5
are from
Hoffleit (1964).
Clusters of galaxies, identifications with more than one plausible
counterpart, and sources for which the identification is uncertain were
excluded.
Also excluded are a few sources for which due to image blending on the POSS
plate no accurate optical position is available. In Fig. 6 (click here)
histograms
with the positional uncertainties derived from our identifications for the
remaining sample of 231 sources are displayed. We plotted the differences of
the positions of the X-ray sources, X,
and the optical counterparts, CP, for declination and right ascension
separately, i.e. 
RA(X-CP) and 
DEC(X-CP), respectively.
Assuming a Gaussian distribution which is suggested by a Kolmogorov-Smirnov
test on a 95% significance level
we derived a 1
error of 
radius in RA
and in DEC each. A small offset between optical and X-ray position in R.A. of
1
to 2
to the East was found. This is due to the motion of the
satellite during the read-out time of the data (cf. Paper I).
Figure 6: Histograms of the positional uncertainties for sources in
areas I, IVac, and Va. The differences in arcsec between X-ray position and
optical counterpart position are plotted separately for RA (solid line) and
DEC (dashed line). 1
, i.e. the 67% error circle, is 
9
. 90% of the counterparts are found within 17
from the
X-ray position in RA as well as in DEC
Figure 7: Cumulative distribution of the distances between X-ray and optical
source position. The resulting 90% error circle is 24
, and the
corresponding 1
uncertainty is 15
The cumulative number of counterparts
within a given distance from the X-ray position is plotted in Fig. 7 (click here).
From this we determined a 1
uncertainty of the distance between
X-ray and optical position of 15
and a 90% error circle of
24
(see Fig. 7 (click here)). The error circle is slightly larger if only
sources with a total number of counts of less than 50 corresponding to
typical count rates of 
ctss
in area I and IVac and
ctss
in area Va are considered. For these sources we
measured a 90% error circle of 27
.
Our estimates for the error circle are, therefore, significantly smaller than
those calculated by the SASS (cf. Paper I) which yielded 90% error circles
on the order of 40
to 45
(see above).
Acknowledgements
The ROSAT project is supported by the Bundesministerium für Bildung und Wissenschaft and the Max-Planck-Gesellschaft. We would like to thank the ROSAT team for observing and processing the ROSAT All-Sky Survey data. We also thank the observers who contributed to the optical observations: J. Alcala, C. Alvarez, H. Bock, H. Bravo, L. Corral, L. de la Cruz, U. Erkens, C. Fendt, Th. Gäng, J. Guichard, M. Kümmel, R. Madejski, A. Marquez, O. Martinez, A. Piceno, A. Porras, F. Ruzicka, Th. Szeifert, J. R. Valdes, F. Valera, G. Vazquez, R. Wichmann, and K. Wilke. We also would like to thank the staff of the Guillermo Haro Observatory for the support during the observations. This work was supported by DARA under grant Verbundforschung 50OR90017.