5.2. Dark Matter and the Structure of the Universe
After my talk at the Caucasus Winter School Zeldovich offered me collaboration in the study of the universe. He was developing a theory of formation of galaxies (the pancake theory, Zeldovich 1970); an alternative whirl theory was suggested by Ozernoy (1971), and a third theory of hierarchical clustering by Peebles (1971). Zeldovich asked for our help in solving the question: Can we find some observational evidence which can be used to discriminate between these theories?
Initially we had no idea how we can help Zeldovich. But soon we remembered our previous experience in the study of galactic populations: kinematical and structural properties of populations hold the memory of their previous evolution and formation (Rootsmäe 1961, Eggen, Lynden-Bell & Sandage 1962). Random velocities of galaxies are of the order of several hundred km/s, thus during the whole lifetime of the Universe galaxies have moved from their place of origin only by about 1 h-1 Mpc (we use in this paper the Hubble constant in the units of H0 = 100 h km s-1 Mpc-1). In other words - if there exist some regularities in the distribution of galaxies, these regularities must reflect the conditions in the Universe during the formation of galaxies. Actually we had already some first results: the study of companion galaxies had shown that dwarf galaxies are located almost solely around giant galaxies and form together with giant galaxies systems of galaxies. In other words - the formation of galaxies occurs in larger units, not in isolation.
Thus we had a leading idea how to solve the problem of galaxy formation: We have to study the distribution of galaxies on larger scales. The three-dimensional distribution of galaxies, groups and clusters of galaxies can be visualised using wedge-diagrams, invented just when we started our study. My collaborator Mihkel Jõeveer prepared relatively thin wedge diagrams in sequence, and plotted in the same diagram galaxies, as well as groups and clusters of galaxies. In these diagrams regularity was clearly seen: isolated galaxies and galaxy systems populated identical regions, and the space between these regions was empty. This picture was quite similar to the distribution of test particles in a numerical simulation of the evolution of the structure of the Universe prepared by Doroshkevich et al. (1980) (preliminary results of this simulation were available already in 1975). In this picture a system of high- and low-density regions was seen: high-density regions form a cellular network which surrounds large under-dense regions.
We reported our results
(Jõeveer &
Einasto 1978)
at the IAU symposium on Large-Scale Structure of the Universe in Tallinn
1977,
the first conference on this topic. The main results were: (1) galaxies,
groups and clusters of galaxies are not randomly distributed but form
chains, converging in superclusters; (2) the space between galaxy
chains contains almost no galaxies and forms holes (voids) of diameter
up to 
70
h-1 Mpc; (3) the whole picture of the distribution of
galaxies and clusters resembles cells of a honeycomb, rather close to
the picture predicted by Zeldovich. The presence of holes (voids) in
the distribution of galaxies was reported also by other groups:
Tully & Fisher
(1978),
Tifft & Gregory
(1978),
and Tarenghi et
al. (1978)
in the Local, Coma and Hercules superclusters, respectively.
Theoretical interpretation of the observed cellular structure was
discussed by
Zeldovich (1978).
Our analysis gave strong support to the Zeldovich pancake scenario. This model was based essentially on the neutrino dominated dark matter model. However, some important differences between the model and observations were detected. First of all, there exists a rarefied population of test particles in voids absent in real data. This was the first indication for the presence of biasing in galaxy formation - there is primordial gas and dark matter in voids, but due to low-density no galaxy formation takes place here (Jõeveer, Einasto & Tago 1978, Einasto, Jõeveer & Saar 1980). The second difference lies in the structure of galaxy systems in high-density regions: in the model large-scale structures (superclusters) have rather diffuse forms, real superclusters consist of multiple intertwined filaments (Zeldovich, Einasto & Shandarin 1982, Oort 1983, see also Bond, Kofman & Pogosyan 1996).