3.1. The morphology-density relation
It was probably Harlow Shapley  in 1926 the first to explicitly refer to the different galaxy content of the Virgo and the Coma cluster, Coma being dominated by ``spheroidal'' galaxy types (7). However, Shapley thought that with increasing resolution many apparently featureless spheroidals would turn out to be real spirals. Ten years after, in The Realm of the Nebulæ, Hubble first hinted at the existence of a morphology-density relation:
``There are some indications of a correlation between characteristic type and compactness, the density of the cluster diminishing as the most frequent type advances along the sequence of classification''
Hubble also noted the ``dominance of late typed among isolated nebulæ in the general field''. The morphology-density relation was immediately regarded as fundamental, to such a point that Tombaugh , in 1937, thought that a galaxy overdensity dominated by spirals could not be a real cluster. In the same year, Tombaugh noted that cluster ellipticals are more centrally concentrated than cluster spirals. In 1942 Zwicky  showed that S0s in Virgo are distributed like ellipticals and unlike spirals.
In 1960 van den Bergh  first noted the existence of a correlation between morphology and local galaxy density. By examining the Ursa Major and Virgo clusters, he noted that
``there is some indication that the nebular population type is related to the surface density of galaxies''
In those years, de Vaucouleurs [131, 132] (see also Abell ) suggested that spirals and ellipticals in Virgo have different distributions simply because they belong to different clusters. The morphology-density relation was thus reduced to a mere projection effect. An even more extreme view was taken by Neyman et al.  who maintained that the observed scarcity of spirals in clusters with respect to the field could be understood as ``a difference in the difficulty of observations''!
In 1965 an extreme case of morphological segregation was discovered. Morgan & Lesh  noted that many clusters are centrally dominated by ``supergiant galaxies'', that they called cDs. These galaxies were shown to live in the densest cluster environment only. Not only are cDs lacking in the field, but also in poor clusters and groups. In fact, the central dominant galaxies of the poor clusters classified by Morgan et al. , were later shown to lack the characteristic extended envelope of cDs (Thuan & Romanishin ).
In the 70's the number of available galaxy redshifts increased considerably, finally allowing a more reliable identification of cluster members. Rood et al.  were then able to identify 16 spirals as members of the Coma cluster. The idea that rich clusters are dominated by ellipticals and S0s was so firmly established that Rood et al.'s was considered a ``striking'' result.
In 1974 Oemler  published his seminal paper The systematic properties of clusters of galaxies. I. Photometry of 15 clusters. He noted that the morphological segregation in clusters depends on the cluster content. The morphology-density relation was interpreted as a relation between the morphological content of a cluster and its compactness. Oemler constructed galaxy number density profiles by type, and noticed a decreasing space density of spirals towards the cluster centres, except in spiral-rich clusters. He also noticed that spirals in cD-clusters have a shallower density profile than ellipticals at large radii. However, he could not notice any difference between the density profiles of S0s and ellipticals.
A year later, Gregory  showed that the fraction of spirals indeed increases with the distance from the Coma cluster centre. He wrote:
``The increase in relative numbers of spiral and irregular galaxies with radial distance seems incontestable. The effect is so strong as to be obvious to the eye on a casual inspection of the Sky Survey''
Melnick & Sargent  confirmed Gregory's finding in other six X-ray bright clusters.
This tendency for ellipticals to be more clustered than spirals was shown by Davis & Geller  not to be restricted to clusters. They applied the 2-point correlation function to the Uppsala catalogue to show that morphological segregation exists on scales up to 6 Mpc. Four years earlier, in 1972, Takase  had already pointed out a colour segregation of galaxies on the scale of the Local Supercluster.
Figure 8. The variation of galaxy population with the mean density of clusters. Solid-line: ellipticals; dashed-line: S0s; dotted-line: spirals. From Oemler (1977).
Figure 9. The fraction of E, S0, and S+I galaxies as a function of the logarithm of the projected density. The upper histogram shows the number distribution of the galaxies over the bins of projected density. From Dressler (1980a).
In 1977 Oemler  wrote that ``density is the physical significant parameter in determining the galaxy population of a cluster.'' Figure 3 of his paper - here reproduced in Fig. 8 - is qualitatively very similar to Figure 4 in the 1980 paper of Dressler  - here reproduced in Fig. 9. Both figures show the fractional variation of spirals, S0s and ellipticals as a function of the cluster density. However, Oemler's density is the mean cluster density, and Dressler's density is the local density around each galaxy. Anyway, Oemler wrote (but did not show) that the same morphology-density relation was also verified individually in clusters dominated by early-type galaxies. The same year, even a spiral-rich cluster (Abell 262) was found to display a ``striking'' morphological segregation (Moss & Dickens ).
Times were mature for Alan Dressler's milestone paper, Galaxy morphology in rich clusters: implications for the formation and evolution of galaxies , published in 1980, and based on the evergreen Catalog of morphological types in 55 rich clusters of galaxies . Dressler pointed out that: i) regular as well as irregular clusters display the same morphology-density relation; ii) it is not the radial distance, but the local density, the basic parameter which determines the morphology mix. Dressler's conclusions are still controversial nowadays (see, e.g. Sanromà & Salvador-Solé ), and it is possible that both global cluster properties and the local galaxy environment may play a role in determining the galaxy morphology .
In the two following years, Bhavsar  and de Souza  extended Dressler's morphology-density relation into the low galaxy density regime, through the analysis of loose groups.
7 It was only in 1923 that Reynolds
out the existence of many ``globular or ovoid'' nebulæ,
distinctly different from spirals.
7 It was only in 1923 that Reynolds  pointed out the existence of many ``globular or ovoid'' nebulæ, distinctly different from spirals. Back.