by Jaan Einasto
Years later when visiting other observatories some astronomers told me that the Tallinn conference was one of the best they ever have attended. Quite recently, I met Hugo van Woerden in Groningen. We discussed problems of common interest, and also the development of the understanding the structure of the Universe. He remembered the Tallinn conference and told me that he and other participants were impressed seeing the enthusiasm that “radiated from eyes of all members of the local organising committee”.
We had wanted to show that Estonia is not the same as the rest of the Soviet Union. We had wanted to show that we have our own culture and traditions, and we belong to the international community. We achieved our goal.
5.2.3 Superclusters, filaments and voids
As the conference progressed, it became clear that the structure of superclusters was its crux. The first speaker who presented new data on the three-dimensional distribution of galaxies was Brent Tully (Tully & Fisher, 1978), a well-known observer who had used radio observations to determine the distances to the galaxies in the Local Supercluster. The core of his presentation was a film of the Local Supercluster. To obtain a spatial image of the supercluster, he used the simple trick of making the image rotate with the use of a computer, which created a threedimensional illusion. This method has later been frequently used by observers of superclusters, as well as by theorists to illustrate the distribution of particles in N-body simulations. The film showed that the Local Supercluster consists of a number of chains of galaxies which branch off from the supercluster’s central cluster in the Virgo constellation. No galaxies could be seen in the space between the chains. This image strongly resembled the one we had acquired of the Perseus supercluster. The structure of the Local Supercluster had been studied for years by the renowned American astronomer Gerard de Vaucouleurs, who also participated in the conference. He implemented supergalactic coordinates which were oriented to follow the Local Supercluster’s denser flat part.
Then it was our turn (Jõeveer & Einasto, 1978). We had been studying the structure of the Perseus–Pisces Supercluster and its surroundings closely, so we had a good understanding of the structure of the Perseus–Pisces and the Local superclusters, as well as of the global network of superclusters and galaxy chains/filaments. The galaxies of the Perseus–Pisces Supercluster form a long chain in which clusters and groups of galaxies are embedded as pearls, see Fig. 5.7. The supercluster’s distance from us is about 50 h−1 Mpc. The main chain of clusters and groups lies nearly perpendicular to the line of sight. Especially characteristic of the chain is its thickness — it is very narrow, as thick as the diameter of the clusters. There are no galaxies either up or downwards of the chain (as seen from our viewpoint), as well as in front of and behind the chain. Thus the chain is an elongated, essentially one-dimensional formation that is surrounded by a void on each side. Its total length is about hundred megaparsecs. At the one end of the chain is the central cluster of the Perseus–Pisces Supercluster (Abell cluster A426, see Fig. 5.7). Further continuation of the supercluster cannot be observed for it is hindered by the absorption effect of the Milky Way. The other end of the chain reaches the next supercluster, which is located somewhat closer to us (Zwicky clusters 481, 487, seen in Fig. 5.7). Another chain of clusters revolve around a large void behind the Perseus Supercluster and form the next system of superclusters behind it at about 130 h−1 Mpc, see Figs. 5.8 and 5.9.
But that was not all of the information about the chain of clusters. Comparison of adjacent slices showed that chains of galaxies and systems of galaxies (groups and clusters) form an almost continuous network. Here and there in the network there are denser regions with more clusters in an aggregate, such gaskets can be considered as superclusters of galaxies. The superclusters in turn are branched; in addition to clusters, there are numerous galaxies which are not randomly scattered but also located as chains. The structure of the Perseus–Pisces supercluster was rather similar to the structure of the Local supercluster, consisting of numerous galaxy filaments, as seen from the movie by Brent Tully, only the richness of knots in chains is different. Chains of the Local supercluster are less rich and contain no clusters, whereas chains of the Perseus–Pisces supercluster contain numerous groups and clusters, in addition to the main cluster A426.
Jõeveer et al. (1978) estimated also the filling factor of the Universe covered by superclusters of galaxies and other filled regions (groups outside superclusters). The data indicated that superclusters fill only about 4 per cent of the total space; the remaining 96 per cent of space forms voids between superclusters. Since voids have been found to exists within superclusters, the filling factor of the Universe with systems of galaxies can be even smaller.
In Figures 5.7, 5.8 and 5.9 we presented the distribution of “near” Zwicky clusters in the sky in the Perseus region in three distance classes. Clusters are numbered according to the Uppsala General Catalogue, Abell clusters are also shown. Cluster contours have been drawn according to Zwicky et al. (1968). Contours of clusters with measured redshifts are drawn by solid lines, clusters with estimated redshifts by dotted lines.
What is especially important is the fact that the Zwicky “near” cluster catalogue is complete — it contains all clusters Zwicky had discovered in this distance interval. Thus there are no selection effects which could distort the distribution. In our first distance interval (3,500–6,500 km s−1) we see the filamentary Perseus— Pisces supercluster, in the second distance interval (6,500–10,000 km s−1) the large void behind this supercluster and clusters surrounding the void, in the third interval (10,000–15,000 km s−1) there is a system of several rich superclusters at mean distance 130 h−1 Mpc, which surrounds the far side of the void of diameter ∼70 h−1 Mpc.
These figures were demonstrated during the Symposium and were included in the Preprint (Jõeveer et al., 1977), but published in the Monthly Notices only in our paper by Einasto et al. (1980a). Later analysis by Einasto et al. (1994b, 1997b, 2001) has shown that most Abell clusters seen in Fig. 5.9 belong to a complex of superclusters at the far side of the void.
A similar analysis was made a few years later by Tago et al. (1984) in the Coma supercluster and its large-scale environment. The distribution of Zwicky clusters of distance class “near” is shown at various redshift intervals in Figs. 1 and 2 of Tago et al. (1984).
The network of superclusters, the chains that form them and the voids in between were comparable to the theoretical model found in the Zeldovich group. To designate this network, we used the term “the cellular structure of the Universe”. In this terminology a “cell” is a low-density region surrounded from all sides by rich clusters and superclusters. Later we used the term “supervoid” to designate supercluster-defined voids (Lindner et al., 1995). Our analysis presented during the Tallinn Symposium showed that cells are not empty — they are crossed by filaments consisting of galaxies and groups of galaxies (poor Zwicky clusters), as seen in Figs. 5.4, 5.5, 5.6, 5.7, 5.8, and 5.9. This means that chains/filaments are of two types: in superclusters they are rich and contain rich clusters and groups, while in cell interiors they are poor and contain only galaxies or poor groups/clusters.
The presence of holes (voids) in the distribution of galaxies was reported also by other groups: by Tifft & Gregory (1978), and Tarenghi et al. (1978) in the Coma and Hercules superclusters, respectively. A few years later a large void in Bootes was discovered by Kirshner et al. (1981), similar in size to the void described above behind the Perseus–Pisces supercluster. The Bootes void in our terminology is the interior of a cell, i.e. a supervoid.
Theoretical interpretation of the observed cellular structure was discussed at the symposium by Zeldovich (1978). As noted by Longair (1978) in his concluding remarks, “the discovery of the filamentary character of the distribution of galaxies, similar to a lace-tablecloth, and the overall cellular picture of the large-scale distribution was the most exciting result presented at this symposium “. These results demonstrated that the
pancake scenario by Zeldovich (1970) has many advantages over other rival scenarios. The term “Large Scale Structure of the Universe” got its present meaning.
However, we noticed also some differences between the observed distribution of galaxies and clusters and the Zeldovich scenario as shown in Fig. 5.3. More on this in Chapter 7.
So far the general perception among astronomers had been that galaxies and clusters are located in space almost randomly. This view was based on the apparent distribution of galaxies and clusters, where voids and sharp features like cluster chains are not visible. Before the Tallinn Symposium I sent the preprint of our results (Jõeveer et al., 1977) among others to Jim Peebles. I got a rapid reply with high-quality photos of the Lick survey and a computer generated 2-D distribution of galaxies using the clustering algorithm by Soneira & Peebles (1978). In the accompanying letter Jim asked: “Can you find filaments in the real and simulated surveys?” In the introduction of the paper Soneira & Peebles (1978) argue: We know that the eye does tend to judge in a biased way — for example, one readily picks out “chains” of points in a uniform random distribution’. Jim and his collaborators have worked very hard to investigate the Lick galaxy counts and other catalogues of galaxies using two-dimensional data and analysis. Our analysis suggests that galaxy filaments are clearly seen only in the three-dimensional distribution of galaxies.
Fig. 5.10 Jim Peebles explaining to Scott Tremain secrets of structure formation, IAU Tallinn Symposium 1977 (author’s archive).
Anyway, the attitude of the astronomical community to our results was rather skeptical, and we had difficulties in publishing the detailed analysis, which was distributed before the Symposium as a preprint. The referee argued that our results are not sufficiently reasoned and demanded that we exclude arguments on the similarity of the present distribution of galaxies to the distribution at the early epoch at the formation of the structure, as well as the need to study the global distribution of galaxies.Also the referee demanded we exclude a detailed theoretical interpretation of the continuous cosmic web which joins all large-scale structures to a single network. We had to make changes in the text many times. Finally the paper was published with an inappreciable title “Spatial distribution of galaxies and of clusters of galaxies in the southern galactic hemisphere” (Jõeveer et al., 1978).
Our experience confirms the opinion of Ernst Öpik (1977), who writes: “From my long experience, both with my own papers, and with those sent to me for refereeing, I have a feeling that the “recognized” journals usually accept without difficulty papers with a middle-of-the-road content, useful contributions to research which already has established itself or accumulations of additional new material. As to pioneering work, papers of this kind often are running the risk of rejection or of excessive curtailment.”
The discovery of voids in the galaxy distribution was recently discussed by Thompson & Gregory (2011). The authors give a fairly complete story of the history of the discovery of voids. I discussed the story with Laird privately via e-mail. Their paper in a refereed journal (Gregory & Thompson, 1978) appeared earlier than ours (Jõeveer et al., 1978). On the other side, we made a preprint (Jõeveer et al., 1977) earlier, and distributed the preprint to participants of the Tallinn symposium. Our paper also arrived at the journal office earlier, but much more time was needed to satisfy all the suggestions of the referee. However the timing is not that important. More important is that we came to the same conclusion on the presence of voids completely independently. To be accurate, the presence of voids in galaxy distribution was first described by Chincarini & Rood (1976).
Gregory & Thompson (1978) used their own observations with limiting magnitude mp < 15 in a 260 degree2 region in the Coma supercluster, the largest possible area around the Coma cluster which could be studied with available equipment in a reasonable timescale. Similar relatively small regions were studied by Tifft & Gregory (1978) and Tarenghi et al. (1978).
What was new in our papers was the complete description of the cosmic web with superclusters, galaxy chains joining superclusters to a connected network, and voids between them. We studied the whole Northern hemisphere from equator up to declination ∼50 degrees. If we exclude regions close to the Milky Way zone of avoidance this makes about one quarter of the whole sky, i.e. about 10 thousand square degrees. Our main goal was to find basic properties of the whole cosmic network. We found that different objects — clusters and groups of galaxies, radio and Markarian galaxies — populate identical regions, and thus can be used as markers of the skeleton of the overall network. For these objects almost complete redshift data were available up to redshift ∼0.05, i.e. distance 150 h−1 Mpc. Redshift samples for galaxies were incomplete, but all available data showed that galaxies have a tendency to cluster in regions marked with clusters and active galaxies. Of course, dwarf galaxies could be distributed more evenly, and fill the voids between galaxy clusters and chains. However, our previous studies of the distribution of dwarf galaxies had shown that dwarf galaxies have a strong tendency to cluster around giant galaxies and to form loose groups around them (Einasto et al., 1974c, 1975b). Thus our approach and that of Gregory & Thompson (1978) were complementary. The advantage of studying the global distribution is that in this case it is possible to investigate properties of the whole cosmic web, not only the environment of superclusters.
To summarise: all essential properties of the cosmic web were discussed by our team at the IAU Tallinn Symposium in 1977 (Jõeveer et al., 1977; Jõeveer & Einasto, 1978), and in subsequent papers by Jõeveer et al. (1978) and Einasto et al. (1980a,b). The basic properties of the web are the following:
(1) The Universe is not structureless but forms the cosmic web as we call it now;
(2) Superclusters consist of chains/filaments of galaxies, groups and clusters of galaxies;
(3) Galaxy and group/cluster filaments form bridges between superclusters;
(4) Chains of galaxies and clusters are essentially one-dimensional, clusters in chains are often elongated along the chain;
(5) The space between galaxy and cluster filaments is almost devoid of galaxies, and form holes/voids;
(6) Voids defined by galaxy filaments have diameters 10… 30 h−1 Mpc, voids defined by rich clusters and superclusters have diameters up to ∼75 h−1 Mpc;
(7) Superclusters occupy about 4% of the total space, the remaining 96% comprises of voids;
(8) The filamentary character of galaxy distribution cannot be formed by clustering of already-formed galaxies — galaxies can form only inside the filaments made of pre-galactic matter, after the dissipation of kinetic energy perpendicular to the filament;
(9) The comparison of the observed structure with results of numerical simulations indicates the presence of non-clustered pre-galactic matter in voids — gravity cannot evacuate voids completely; galaxies form only in regions of enhanced density (see Chapter 7 for discussion of this property of the web);
(10) Observed large-scale distribution of matter is the remnant of singularities in the initial perturbation field.
However, further development of the cosmic web concept was influenced by the East–West controversy during the Cold War. Fairall (1998) characterises the reception of the concept of the cosmic web by the Tartu team with the following words: “News of the claim travelled widely, but converts outside the Soviet Union were few; the idea seemed to overthrow all that was understood in the West about clustering. It was seen to support the Soviet theoretical cosmological view — in particular that of Zeldovich and colleagues in Moscow — involving ‘pancaking ’, rather than the American views of galaxy formation, at a time when ‘cold war ‘ rivalry flourished.”
The reception of the concept of the cosmic web was partly hindered by the suspicion that our data are incomplete and thus biased, as noted by Joe Silk in his interview in Lightman & Braver (1992). This shows that readers have not noticed one of the basic aspects of our study: to find the skeleton of the cosmic web we used data not o
nly for galaxies, but also for near clusters of galaxies and active galaxies — these data were complete. Just the combination of various markers of the cosmic web allowed us to find almost all essential features of the web already in the late 1970’s, much earlier than similar results were obtained from complete redshift data covering large contiguous regions on sky.
Soon we understood that the structure of the cosmic web contains information not only on the scenario of galaxy formation, but also on the nature of the dark matter — the basic constituent of the matter in the Universe. Further development of our understanding of the nature of the dark matter and of the structure of the cosmic web occurred hand in hand. It shall be described in the following sections.
5.3 Tartu Observatory in the early 1980’s
5.3.1 Southern base of Tartu Observatory
All our results on the distribution of galaxies and the structure of the Universe have been achieved by analysing observations that had been made elsewhere. In the 1970’s we attempted to measure radial velocities of galaxies using our 1.5-m telescope and later, in cooperation with the Byurakan Observatory, using their 2.6-m telescope. Results of these radial velocity measurements were published by Vennik et al. (1982); Vennik & Kaasik (1982). In Tartu Observatory we got redshifts for 62 galaxies, in Byurakan for 34 galaxies, located in groups of galaxies. However, the climate both in Estonia and Armenia does not allow a statistically adequate sample of galaxies to be observed.