We have assembled a large data-set of 613 galaxy redshifts in the Coma
cluster, the largest presently available for a cluster of galaxies. W
e have defined a sample of cluster members complete to b(26.5) = 20.0,
using a membership criterion based on the galaxy velocity, when avail
able, or on the galaxy magnitude and colour, otherwise. Such a data se
t allows us to define nearly complete samples within a region of 1 h(-
1) Mpc radius, with a sufficient number of galaxies per sample to make
statistical analyses possible. Using this sample and the ROSAT PSPC X
-ray image of the cluster, we have re-analyzed the structure and kinem
atics of Coma, by applying the wavelet and adaptive kernel techniques.
A striking coincidence of features is found in the distributions of g
alaxies and hot intracluster gas. The two central dominant galaxies, N
GC 4874 and NGC 4889, are surrounded by two galaxy groups, mostly popu
lated with galaxies brighter than b(26.5) = 17 and well separated in v
elocity space. On the contrary, the fainter galaxies tend to form a si
ngle smooth structure with a central peak coinciding in position with
a secondary peak detected in X-rays, and located between the two domin
ant galaxies; we suggest to identify this structure with the main body
of the Coma cluster. A continuous velocity gradient is found in the c
entral distribution of these faint galaxies, a probable signature of t
idal interactions rather than rotation. There is evidence for a bound
population of bright galaxies around other brightest cluster members.
Altogether, the Coma cluster structure seems to be better traced by th
e faint galaxy population, the bright galaxies being located in subclu
sters. We discuss this evidence in terms of an ongoing accretion of gr
oups onto the cluster.