5 Conclusions

We analysed three clusters of galaxies with low to intermediate X-ray luminosities which show an irregular appearance in the ROSAT All-Sky Survey. As the confusion with fore- and background sources is increasingly critical with decreasing flux and increasing substructure of the cluster, we test the limitations of the RASS. With only three clusters, of course, this cannot be a statistical study, but it is meant to be a "worst-case'' study. We would like to stress that for higher fluxes source confusion is less important and that for clusters with these fluxes the RASS results are very reliable.

The results of the follow-up observations - X-ray observations with the ROSAT/HRI and optical spectroscopic observations - are summarized in Table 5. The ROSAT/HRI observations, which have a much better spatial resolution than the RASS, revealed in two of the three cases a different morphology than seen in the RASS. The reasons are point sources which could not be resolved in RASS and were therefore confused with the cluster emission. In one of the three clusters the countrate measurement is affected by the point sources, in the two other cases the countrate measurements of the RASS are reliable. The results for the three clusters are the following:

• Abell 901: With a flux of $f_{\rm X}(0.1-2.4$ keV) = $3.0\pm0.1~10^{-12}$ erg/cm²/s this cluster is the faintest one in the sample. Both the measurement of the countrate and the determination of the morphology provide difficulties in the RASS. Several discrete sources cannot be distinguished from the cluster emission in the RASS. With the RASS information only it was concluded that this cluster consists of two subclumps (Ebeling et al. 1996), while the true cluster X-ray emission is very compact ( $r_{\rm c}=22$ kpc). The countrate would be overestimated by at least 80% if the discrete sources cannot be separated from the cluster;

• Abell 1437: This cluster has the highest flux of all the clusters in the sample $f_{\rm X}(0.1-2.4$ keV) = $1.04\pm0.03~10^{-11}$ erg/cm²/s. With such a high flux the RASS results are very reliable. There is no problem with the countrate or morphology determination. The point source in the NE of the cluster can easily be distinguished in the RASS;

• Abell 3570: This cluster has several point sources superposed on the cluster emission. Although the countrate determination is not affected by these sources (their countrate is only 1% of the cluster countrate) the true morphology is very regular in contrast to the impression from the RASS. Optical measurements of the cluster galaxies give a very small velocity dispersion confirming the picture of a relaxed cluster.

We conclude that for clusters with fluxes less than a few times 10^-12 erg/cm²/s, which have at the same time an irregular morphology, the confusion with fore- and background sources can be a problem in the RASS. So morphological analyses of RASS clusters tend to overestimate the fraction of clusters with substructure.

A new X-ray all-sky survey with a potential second ABRIXAS (Friedrich et al. 1996) mission would have several advantages to push this source confusion limit down.

• The spatial resolution of ABRIXAS is about 2.5 times better than the resolution of the RASS, which makes it easier to distinguish between point and extended sources;

• The energy range of ABRIXAS (0.5-12 keV) is much better suited for clusters than the ROSAT range (0.1-2.4 keV). Although both surveys have about the same sensitivity at 1 keV, ABRIXAS with its harder energy range would detect 3-5 times more photons from a standard cluster;

• The spectral resolution of ABRIXAS is better than that of the ROSAT/PSPC. Together with the wider energy range of ABRIXAS this provides an improved possibility to separate hard cluster emission from soft foreground sources.

Acknowledgements

I thank Chris Collins for introducing me into the secrets of optical spectra, Hans Böhringer for making the RASS images of the three clusters available to me, Peter Friedrich for providing the ABRIXAS numbers, and Phil James and Joachim Wambsganss for carefully reading the manuscript. It is a pleasure to thank Carlo Izzo for his most helpful EXSAS support. I acknowledge gratefully the hospitality of the Institut d'Estudis Espacials de Catalunya in Barcelona. During the stay there I was supported by the TMR grant ERB-FMGE CT95 0062 by CESCA-CEPBA.