8 Further tests of the sample completeness

To further test the completeness of the REFLEX sample we have conducted two additional searches for clusters. The first search is based on the X-ray source extent and the second on a systematic search for X-ray emission from Abell clusters.

In the search for clusters among the extended RASS sources we have inspected all sources in the flux limited sample ( $F_{\rm X} \ge 3~10^{-12}$ ergs^-1 cm^-2) which feature a KS-probability less than 0.01 of being point-like and which are not already included in the REFLEX sample. In total 48 additional extended sources are found (after removal of strange detections at exposure edges and fragments of larger clusters already included in REFLEX). 35 of these sources are identified with bright stars and QSO and we notice that they are often borderline cases concerning the extent significance; another fraction of these sources feature an extent in the analysis without deblending because they are close pairs of point sources with identifications other than clusters. The remaining objects for which a cluster identification cannot be ruled out are in total 13 X-ray sources: 8 certain clusters, 2 good looking cluster candidates and 3 fields with no indication for a galaxy cluster and no obvious other identification. We are planning further deeper imaging for the latter sources. Thus we have found about 10 objects in this search which have been missed in the REFLEX compilation. The above result can also be used for another interesting and useful statistic. The 35 partly spuriously extended sources among the non-cluster candidates (plus the 28 extended non-cluster sources mentioned in Sect. 10) if compared to an initial sample of 1050 sources (above the REFLEX cut with extended cluster sources subtracted) implies a failure rate of flagging non-extended X-ray sources erroneously as extended of less than 6%.

The Abell and ACO catalogues (Abell 1958; Abell et al. 1989) contain about 5 times as many objects as the REFLEX sample in the study area. Even so we do not expect a very close match of the two samples, since e.g. the correlation of X-ray luminosity and optical richness is quite weak (see e.g. Ebeling et al. 1993), the large overabundance of Abell clusters provides a good check regarding problems in the recognition of clusters by the galaxy count technique based on COSMOS. To search systematically for X-ray emission from all ACO and ACO supplementary clusters we run the GCA algorithm on all ACO positions allowing for a recentering of the method within a radius of 10 arcmin of the input position. We find only one ACO supplementary cluster that was not flagged by the galaxy counts and should be included in REFLEX given its GCA flux. This cluster was already found in the above discussed additional cluster search at the positions of the extended RASS X-ray sources. It happens that this cluster is actually close to the boundary to the Large Magellanic Cloud which might explain the deficiency in counted galaxies at this position.

Since the search for X-ray emission from ACO clusters is independent of the previous source detection in the RASS II primary source list, we are not only testing the completeness of the cluster finding by the optical galaxy counts but also the source detection in the RASS II standard analysis (Voges et al. 1999). Since we find no ACO cluster missing in REFLEX due to its non-detection in RASS II, we can conclude that missing of sources in RASS II is not a significant problem for the completeness of the REFLEX sample. Such completeness of the primary source detection will be studied further by simulations of the source detection efficiency in the RASS data.

In summary, from the available material we find a missing fraction of clusters of about 2-3% in REFLEX which can be recovered as described in this section. This small fraction is still well consistent with the internal estimate of a completeness of over 90% and further supports the quality of the REFLEX sample. Note that the additional cluster detections are not integrated into the REFLEX sample to conserve its homogeneity but will be listed as REFLEX supplementary clusters in forthcoming catalogue publications.