2.3. Superclusters and voids
In his milestone paper, Abell  also demonstrated the existence of ``clusters of clusters'' in 3 dimensions. Abell used his magnitude-based cluster distance estimates to establish that the average size of superclusters is 60 Mpc. He rejected Zwicky's hypothesis of IC dust by showing that regions of the sky devoid of intermediate-distance clusters were nevertheless occupied by even more distant clusters. Ten years after, Reaves  was able to set an upper limit of 0.1 magnitudes to the extinction by IC dust, based on the colour vs. redshift relation for galaxies in cluster fields. Despite Abell's and Reaves' results, Bogart & Wagoner  in 1973 still invoked IC dust as the origin of an apparent cluster-cluster anti-correlation.
In 1962 Abell  published the first list of (seventeen) superclusters. He noted that the existence of superclusters was to be taken into account when estimating the probability of chance projection effects in a cluster catalogue, thus anticipating the ideas of Lucey . A few years later, Abell & Seligman  showed that superclusters could be easily identified even in Zwicky's CGCG .
A step further towards establishing the reality and properties of the Local Supercluster, was done by de Vaucouleurs . He considered the distribution of 55 nearby groups. By noting that 85 % of all nearby galaxies are in groups, he suggested that superclusters may well overlap and fill all the space available. He correctly argued that the observational samples had not yet reached to the distance of homogeneity, thus making it meaningless any attempt to estimate the mean density of the Universe. The concept of the Large Scale Structure of the Universe was taking his first steps.
Despite this observational progress, the reality of superclusters remained an open issue. Peebles and collaborators published papers arguing both against  and in favour of the existence  of superclusters. Peebles' final word came in 1974, with the development of a mathematical tool that was to stay with cosmologists ever since: the covariance function . By showing that the covariance function is a simple power law over a very large distance range, he concluded that there was no physical division between groups and clusters, nor between clusters and superclusters.
Zwicky continued to reject all evidences in favour of the existence of superclusters. He thought that IC dust could account for irregularities of the clusters distribution. Zwicky's hypothesis was finally falsified by Reaves  in 1974. Reaves showed that intermediate-distance clusters are less often seen behind nearby clusters than very distant ones. Correctly, he attributed this to the difficulty of distinguishing clusters in projection when they are not well separated along the line of sight, and the two cluster luminosity functions peak at a similar magnitude.
Fritz Zwicky did not live long enough to read Reaves' paper. He died on Feb. 8th 1974, just a few days before his 76th birthday.
Figure 5. The wedge diagram of the Coma supercluster; crosses indicate galaxies that would be too faint to be detected if they were at the distance of the Coma cluster - from Gregory & Thompson (1978).
After Zwicky's death the reality of superclusters was no longer questioned. A major breakthrough in this topic came with the extensive redshift surveys of Chincarini, Gregory, Rood, Tarenghi, Thompson & Tifft [459, 195, 454, 449, 109, 448, 196], that drew the 3-dimensional structures of the Coma - see Fig. 5 - Hercules, Hydra-Centaurus, Perseus and Pisces superclusters. Cluster-connecting filaments and voids were identified. The emerging picture was thus summarized by Abell :
``The picture that suggests itself is that of a large inhomogeneity or region of space containing galaxies, groups, and clusters, in which what is commonly called the Coma cluster is simply a dense concentration, rather like an urban center in a large metropolitan area''
In 1978 Jôeveer et al.  described Perseus and other eight superclusters, and noted that the majority of clusters of galaxies form chains. Einasto et al.  pointed out that the large scale structure of the Universe resembles cells, with galaxies and galaxy clusters concentrated towards cell walls, whereas the spatial density of galaxies inside cells is very low. In 1981 Kirshner et al.  found the million Mpc3 Boötes void, that Bahcall & Soneira  showed to be associated with the Hercules supercluster and the CorBor extension.
Numerical simulations were keeping abreast of observations: in 1979 Aarseth et al.  were able to produce 3-dimensional plots of the galaxy distribution where the recently discovered huge voids were quite evident (6).
6 In the discussion following Aarseth's
Peebles referred to Aarseth's plots as ``propaganda films'' and
deemed it ``very dangerous to compare
them too closely to the real Universe''.
6 In the discussion following Aarseth's talk , Peebles referred to Aarseth's plots as ``propaganda films'' and deemed it ``very dangerous to compare them too closely to the real Universe''.