2 Sample selection

Our sample is selected from RASS-BSC according to the following criteria:

1.

Declination $\delta \geq 3\hbox{$^\circ$}$.

2.

Galactic latitude $\vert b\vert \geq 20\hbox{$^\circ$}$.

3.

Optical counterparts within a circle with radius $r=r\hbox{$^\prime$}+5\hbox{$^{\prime\prime}$}$,where $r\hbox{$^\prime$}$ is the RASS position error given by Voges et al. (1996b).

4.

Optical counterparts with R magnitudes between 13.5 and 16.5, where R magnitude was derived from USNO-A1.0.

5.

${ \log C \geq - 0.4~R + 4.9}$.
where C is the X-ray count rate of RASS source and R is the R magnitude of possible counterpart. The results of the EMSS (Stocke et al. 1991) have shown that different classes of X-ray sources represent different narrow ranges in the X-ray-to-optical flux ratios. These bounds mean the X-ray sources with Galactic and extragalactic counterparts can be separated at high confidence level prior to any optical spectroscopy (Maccacaro et al. 1988). However, X-ray flux is difficult to evaluate before optical identification. In spite of the absorption by the column density $N_{\rm H}$ of cold material X-ray count rate is roughly proportional to X-ray flux, so X-ray flux in the X-ray-to-optical flux ratio criterion can be replaced by X-ray count rate, i.e., ${ \log (f_{\rm X}/f_{\rm O}) \propto \log C + 0.4\,R + {\rm constant}}$, where we use the R magnitude flux to represent the optical flux. According to our statistical analysis of known RASS-BSC X-ray sources, we found there is an apparent gap between Galactic stars and extragalactic objects. Emission line AGNs concentrate on the region: ${ \log C \geq - 0.4\,R + 4.9}$. Thus choosing ${ \log C \geq - 0.4\,R + 4.9}$ as preselecting AGNs criterion would be efficient. Details of the criterion construction have been described in Cao et al. (1998);

6.

No association with objects in the Galactic and extragalactic catalogues compiled in the SIMBAD and NED database.