5. Overall source statistics
5.1. Completeness of identification and repartition by types
The final list of proposed optical counterparts including the positional and Ca II derived probabilities of identification are listed in Tables 11 and 12. We show in Tables 3 and 4 the statistics of source identifications in the 'full' area for several limiting count rates. For sources detected with ML 10, the X-ray completeness level is 0.02 cts s-1 (see Sect. 9) and at this limit we have optically identified 88% of the sources. Active coronae constitute the overwhelming majority of our sample. The contribution of white dwarfs reduces the stellar fraction at high count rates and the incompleteness of the optical identification again decreases the fraction of active coronae at low count rates. At the level of 0.03-0.025 cts s-1 which is about the median all-sky survey sensitivity we have identified over 95% of the sources and active coronae account for about 85% of the total population in our low galactic latitude field. A high fraction of active coronae was also found in the Einstein Galactic Plane Survey ( 46%; Hertz & Grindlay 1984, 1988) and in a ROSAT sample area in Perseus ( 48%; Motch et al. 1991). Our record beating percentage of active coronae in Cygnus obviously reflects the location at b = of our area. A similar test region in the Taurus Constellation, away from the main star forming region (b ; l ) also contains 77% of active stars with an optically identified extragalactic population of 13% (Guillout 1996). At higher galactic latitudes, the fraction of active coronae identified in the RASS falls to 42% (Zickgraf et al. 1996) and may reach values as low as 10% at the north galactic pole (Zickgraf 1996). It is thus amazing that in spite of the apparently nearly constant density of ROSAT survey sources with galactic latitude ( 1.5 deg-2 ; Voges 1992) their nature changes from an extragalactic dominated to star dominated population while moving to lower galactic latitudes.
Table 3. Source identification statistics in the 'full' area (all ML). Figures are number of sources
Table 4. Source identification statistics in the 'full' area. Figures are in percents
5.2. Positions and 90% confidence radii
An accurate knowledge of the ROSAT error circle is of utmost importance for guiding the optical observations in the crowded fields often encountered at low galactic latitude. An illustrative example is probably our discovery of the WD+Me binary RX J2130.3+4709 (index 84). Spectroscopy of the bright (V = 8.45) G0V star HD 204906 failed to reveal convincing Ca II emission and furthermore the star was clearly outside the r90 radius. This led us to investigate other candidate stars closer to the X-ray position and discover this interesting system.
The errors on ROSAT X-ray positions have two different origins. First, the uncertainty with which the centroid of the X-ray image is positioned on the pixel grid by the maximum likelihood source detection algorithm. Second the error in the knowledge of the satellite attitude for each photon collected in scan mode. Early analysis led to the conclusion that the error on the position was dominated by the ROSAT attitude bore sight accuracy and that assuming a Gaussian two-dimensional distribution the final uncertainty could be written as:
where is the maximum likelihood error and where , the attitude uncertainty was estimated to be from a subset of X-ray binaries with accurate positions (Motch et al. 1996b). Applying this estimate to the sources in our survey field yields a mean r90 in the range of to slightly depending on the maximum likelihood of detection (see Table 5). The histogram of the distance between X-ray and optical positions expressed in units of the 90% confidence radius (see Fig. 10) is compatible with a normal Rayleigh distribution. The fact that only 7% of all 86 proposed optical counterparts are located outside r90 from the ROSAT position suggests that we may have slightly overestimated the satellite attitude error and that the actual is close to . Adding the 14 less secure candidates identified at the 95% confidence level only does not change this conclusion. However, the identification strategy may have favoured optical counterparts located close to the X-ray position. The positions extracted from SIMBAD are based on 1950 coordinates and are not corrected for proper motion. Also the "Quick V" plate collection on which are based the northern GSC coordinates was obtained in the 1982-1984 interval and these positions may be altered to some extent by the unknown proper motion. Therefore, a small fraction of the scatter in the distance between X-ray and optical positions may be due the lack of correction for proper motion.
Table 5. 90% confidence radii for various maximum likelihood
5.3. X-ray hardness ratios
In principle hardness ratios are powerful tools to select and guide the identification of ROSAT X-ray sources (see e.g. Motch 1992). However, the use of X-ray colours is essentially limited to the brightest sources. With a mean exposure time of the order of 800 s useful information may be extracted for the sources having count rates larger than 0.02 cts s-1. We show in Fig. 11 the distribution of a subset of 62 sources with count rates larger than 0.012 cts s-1, in the HR1/ HR2 diagram. Hardness ratios 1 and 2 have here energy boundaries as used in the latest SASS versions:
where (A-B) is the raw background corrected source count rate in the A-B energy range expressed in keV. Identified active coronae span a wide range of X-ray colours consistent with the variety of temperatures encountered in stars (e.g. Schmitt et al. 1990). White dwarfs populate the soft region of the diagram while our only identified AGN exhibits the hard X-ray colours expected from a heavily absorbed source. The unidentified source RX J2133.3+4726 (index 35) is located on the sky close to the identified AGN and displays hard X-ray colours (HR1 = 0.96 0.27; HR2 = 0.62 0.16). Deep optical searches in this error circle indeed rule out identification with active coronae (Motch et al. 1996a). This suggests a remote luminous source possibly of extragalactic nature. The O8Ve star HR 8154 also exhibits a soft HR2 and hard HR1 which probably originates from a substantial interstellar absorption toward this remote (d 640 pc) source in CYG OB7. Active coronae detected at large distance may have harder HR1 than closer ones (see Fig. 12). However, this effect could either be due to enhanced photoelectric absorption towards most distant objects and / or to intrinsically higher temperatures usually observed for the most luminous active coronae which are obviously the ones we detect at the largest distances (Schmitt et al. 1990).
5.4. X-ray variability
Owing to the snapshot nature of survey observations for which we have at best a 32 s long integration every 96 min only limited information on time variability is available. However, active coronae are known to exhibit large X-ray flares and one may wonder whether a sizeable fraction of our sources is detected thanks to these fast events. Visual inspection of the light curves produced in the broad (0.1-2.0 keV) and hard (0.5-2.0 keV) X-ray band reveals that most active coronae were detected during all satellite orbits and that the flaring activity is in general not prominent. Our only conspicuous flare originates from the Me star RX J2104.1+4912 (index 7; see Fig. 13). In the absence of this particular event, the mean count rate from the source would have been 0.062 cts s-1 instead of the 0.13 cts s-1. Accordingly, we conclude that the effect of flares is not important in our sample and that there is no large bias in favour of the preferential detection of flare stars.
© European Southern Observatory (ESO) 1997
Online publication: July 8, 1998