Clusters of galaxies are ideal probes of large -scale structure as they are rare objects;
large volumes of the Universe can be mapped with relatively small (when compared to
galaxies) numbers of clusters. Clusters are also bright, they are among the most luminous
X -ray sources in the sky. The X -ray emission originates from hot, diffuse, intracluster gas.
The distribution of the gas density follows that of the underlying (dark) mass potential.
Consequently the cluster X -ray emission is concentrated towards the central core region
of the cluster. In the past it was not possible to make large X -ray selected samples of
clusters because the X -ray data suffered from either poor spatial resolution or inadequate
areal coverage. Therefore, cluster samples have traditionally been developed by searching
for projected galaxy enhancements in photographic or CCD surveys. However, such
optically selected cluster samples have been shown to suffer from projection effects which
bias their clustering properties and limit their use as diagnostic tools. The work described
in this thesis concerns the development of the SGP RASS Cluster Sample (SRCS), the
first truly X -ray selected sample of clusters, and its application to a quantitative study
of large -scale structure.
The SRCS has been constructed from data acquired during the ROSAT All Sky
Survey (RASS), where ROSAT is the German /UK /US X -ray satellite that was launched
in 1990. The RASS, completed in 1992, was the first all sky survey made with an imaging
X -ray telescope. Approximately 50,000 sources were detected during the RASS of which
roughly 10% are clusters of galaxies. We have selected 345 cluster candidates from the
RASS, to a limiting flux of fx 1 X 10 -12 erg s -1 cm-2, from a 28000° region centered
around the South Galactic Pole (SGP). These candidates have been selected using both
X -ray data from the RASS and digitised optical data from the COSMOS /UKST Object
Catalogue (CUOC). Through a combination of optical spectroscopic data from our own
observations and from the literature, we have been able to identify 154 of these candidates
as clusters with redshifts. An additional 36 candidates have been identified with non - cluster X -ray sources that were contaminating the candidate list. The 154 RASS clusters
in the SRCS constitute the largest X -ray selected cluster sample to date.
Our experience has provided us with a unique insight that will aid future searches for
clusters in the RASS database. We have been found that a significant fraction, 20 %, of
RASS clusters are not Abell optical clusters (Abell 1958, Abell et al. 1989). We have also
shown that 40% of RASSclusters were not flagged as extended sources by the Standard
Software Analysis System (SASS). We conclude, therefore, that it is not possible to
make comprehensive identifications of RASS clusters by relying solely either on the
Abell catalogue or on existing SASS extent information. We also recommend that all
candidates with a single galaxy in the X -ray error circle be followed up spectroscopically
irrespective of whether the galaxy lies in a cluster with a known redshift. (This is due to
the enhanced probability that such candidates are in fact X -ray bright AGN.) We note
that planned changes to SASS may provide the means to identify contaminating objects
without optical follow -up, by improving extent determinations. Such changes will also
allow us to address one of the major drawbacks of the work presented in this thesis; that
we have had to use unreliable source count rates.
Using the SRCS, we have been able to make the first determination of the spatial
correlation function, ¿cc, for an X -ray selected cluster sample. A reliable measurement
of X ray was long overdue, as it can provide important constraints on theories of large - scale structure formation. Constraints from ¿cc ° are more robust than those derived
from measurements of optical cluster clustering, C ,zcal, because X -ray selected cluster
samples do not suffer from the projection biases mentioned earlier. We hope to settle the
controversy over the value of the cluster correlation length, ro, using the SRCS. (Where
ro is the separation at which ¿cc = 1.)
We have found that the clustering patterns of the SRCS are similar, on all scales,
to those of automated digitised optical cluster samples. In contrast, when we compare
the SRCS to samples of Abell clusters, clusters which were selected by eye by Abell and
co- workers from photographic plates, we measure a smaller correlation length (r ó "Y
16 h -1 Mpc cf. ró bell Z 20 h -1 Mpc) and less power on large (r Z 40 h-1 Mpc) scales.
We attribute this difference to artificial clustering in the Abell catalogue, as evidenced
by our comparisons of the two - dimensional clustering in our sample with that in Abell
samples. The lack of anisotropy seen in the SRCS correlation function provides strong
support for the claims of Sutherland 1988 that some of the clustering power seen in
Abell cluster samples is due to projection effects. Alternative suggestions proposed to
explain the line -of -sight anisotropy in the Abell catalogue, e.g. large peculiar velocities
or supercluster elongations, are much less tangible in light of our results.
We have made comparisons between the SRCS correlation function and state of the
art theoretical predictions. Our data seem to rule out a low density, positive cosmological
constant, CDM model. However, they provide inadequate constraints to be able to
distinguish between other, subtly different, models, e.g. standard CDM, mixed CDM
and tilted CDM models. Larger X -ray cluster samples would be required to do so and it
is planned to extend the project to produce a larger, more complete, SRCS in the near
future.