1999A&A...341..371D

Analysis of the data obtained in the ESO Nearby Abell Cluster Survey (ENACS) has shown that the space distribution and kinematics of galaxies with detectable emission lines in their spectra differ significantly from those of galaxies without emission lines. This result, and details of the kinematics, were considered as support for the idea that at least the spirals with emission lines are on orbits that are not isotropic. This might indicate that this subset of late-type galaxies either has `first approach'-orbits towards the dense core of their respective clusters, or has orbits that `avoid' the core. The galaxies with emission lines are essentially all late-type galaxies. On the other hand, the emission-line galaxies represent only about a third of the late-type galaxies, the majority of which do not show detectable emission lines. The galaxies without emission lines are therefore a mix of early- and late-type galaxies. In this paper we attempt to separate early- and late-type galaxies, and we study possible differences in distribution and kinematics of the two galaxy classes. For only about 10% of the galaxies in the ENACS, the morphology is known from imaging. Here, we describe our classification on the basis of the ENACS spectrum. The significant information in each spectrum is compressed into 15 Principal Components, which are used as input for an Artificial Neural Network. The latter is `trained' with 150 of the 270 galaxies for which a morphological type is available from Dressler, and subsequently used to classify each galaxy. This yields a classification for two-thirds of the ENACS galaxies. The Artificial Neural Network has two output classes: early-type (E+S0) and late-type (S+I) galaxies. We do not distinguish E and S0 galaxies, because these cannot be separated very robustly on the basis of the spectrum. The success rate of the classification is estimated from the sample of 120 galaxies with Dressler morphologies which were not used to train the ANN. The success rate is higher for early-type than for late-type galaxies (78±6% vs. 63±6%). The weighted average success rate, irrespective of type, is 73±4%. The success rate is somewhat larger for the training set, and highest for the galaxies with emission lines. Of the 3798 galaxies that were classified from their spectrum 57±7% are of early type, and 43±7% of late type. Using a subset of these 3798 galaxies, we constructed a composite cluster of 2594 galaxies, 399 of which have emission lines and are therefore almost exclusively spirals and irregulars. The kinematics and spatial distribution of the late-type galaxies without emission lines resemble much more those of the early-type galaxies than those of the late-type galaxies with emission lines. Yet, the late-type galaxies without emission lines may have a somewhat larger velocity dispersion and a slightly less centrally concentrated distribution than the early-type galaxies. Only the late-type galaxies with emission lines appear to have a considerably larger global velocity dispersion and a much less concentrated projected density profile than the other galaxies. Thus, the suggestion of fairly radial, and possibly `first approach' orbits applies only to spirals with emission lines. The early-type galaxies with emission lines (among which the AGN), may also have a large velocity dispersion and be concentrated towards the cluster centre.