1. Fin-de-Siècle Cosmologies

Astronomers and physicists in the 19th century rarely dealt with questions of cosmology and cosmogony, which were widely considered to belong to philosophical speculations (Merleau-Ponty, 1983). Yet there was an interest in problems related to the structure and evolution of the universe at large, such as whether cosmic space and its material content were finite or infinite. What became known as Olbers’s paradox was formulated by the German astronomer Heinrich Wilhelm Olbers in 1826 but has a history that goes much farther back in time (Harrison, 1987; Jaki, 1969). The essence of this so-called paradox is that in an infinite universe uniformly populated with stars, the sky at night should be as bright as on a sunny day or even brighter. The explanation favored by Olbers and most other astronomers at the time was to assume that space is filled with a light-absorbing medium. It was generally agreed that space is infinite, and in the 19th century, astronomers did not generally regard the darkness of the night sky as a serious problem. It only became so in the early part of the 20th century when it turned out that interstellar space was much more transparent than Olbers and his contemporaries had assumed.

The question of the size of the stellar universe also turned up in connection with the paradigmatic belief that in the end, it was governed by Newton’s law of gravitation. In works of 1895 and 1896, the prominent German astronomer Hugo von Seeliger proved that an infinite universe with stars distributed approximately uniformly could not be brought into agreement with Newton’s law (Norton, 1999). Seeliger did not conclude that the universe was therefore finite—which would lead to other problems—but instead suggested a slight modification of the law at very large distances. While Newton’s law states that the force between two masses is attractive and varies inversely with the square of their distance, Seeliger added an exponential attenuation factor corresponding to a long-range repulsive force. In this way, he could save the cherished infinite stellar universe and still remain on the safe ground of classical mechanics.

The problem highlighted by Seeliger came to be known as the “gravitational paradox,” a kind of mechanical analog to Olbers’s optical paradox. Without relying on Seeliger’s ad hoc hypothesis of a repulsive cosmic force, the problem was also considered by other astronomers in the early 20th century. The paradox or problem assumed a homogeneous distribution of matter, an assumption that some astronomers questioned. Carl Charlier, a leading Swedish astronomer, developed in 1908 a so-called hierarchic model of the universe that allowed it to be finite and yet in accordance with Newtonian mechanics (Jaki, 1969; Norton, 1999). Thus, by making more or less artificial assumptions, it was possible to argue in both favor and disfavor of a finite stellar universe. However, most astronomers and physicists agreed that space itself was infinite if not necessarily filled throughout with matter.

As regards the temporal extension of the universe, the consensus view was either to ignore the issue or to defend an eternal universe as the only reasonable possibility. As Charlier stated in a paper of 1896, “A finite time is a contradiction. . . . An infinite time may be difficult to conceive, but it is not contradictory” (Charlier, 1896, p. 481).

The discussions mentioned so far took it for granted that space was flat or Euclidean. Although non-Euclidean geometries with curved space had been known since the 1830s, it took a generation or more before these ideas made an impact on astronomers’ conception of space. As a few mathematicians realized in the 1860s, there are three and only three geometries with constant curvature: flat space (zero curvature), spherical space (positive curvature), and hyperbolic space (negative curvature). In terms of the dimensionless curvature constant k, the three possibilities were given by k = 0, k = +1, and k = −1, respectively. Which of these possibilities corresponded to the real physical space in which people live?

The first astronomer to consider the question in a cosmological context was the German astrophysicist Karl Friedrich Zöllner, who in a little-noticed work of 1872 focused on the case of constant positive curvature (k = +1) first considered by the mathematician Bernard Riemann in the 1850s. As Zöllner pointed out, if cosmic space were positively curved in analogy with the two-dimensional space of the surface of a sphere, then space and all its material content would be finite and yet there would be no limit to the universe (Kragh, 2012b). Since there were only a finite number of stars in such a model universe, it offered a novel solution to Olbers’s paradox. However, neither Zöllner himself nor other scientists followed up on his brilliant argument, which left no trace in the further development of cosmology.

Although a few astronomers in the late 19th century considered the possibility of a non-Euclidean universe, it was more as an academic exercise than because they believed that the universe might actually be curved. Leading scientists such as the American astronomer Simon Newcomb and the French mathematician and physicist Henri Poincaré maintained that although space might be curved, there were neither observational nor theoretical reasons to believe that it was. According to Poincaré’s conventionalist view of science, the geometry of space was nothing but a convention and not something that could be determined objectively.

The first astronomer since Zöllner who seriously examined the geometrical structure of space was young Karl Schwarzschild, who in 1900 systematically compared astronomical observations with expectations based on the hypothesis of curved space (Kragh, 2012a). However, the data available to Schwarzschild did not allow a definite conclusion except that they indicated a lower bound for the radius of curvature corresponding to either a spherical (closed) or a hyperbolic (open) universe. In the first case, he concluded that the radius of the universe must be greater than 1,600 light years. Although Schwarzschild’s careful investigation was known to contemporary astronomers, it did not disturb the consensus view that in all likelihood, cosmic space was flat and consequently infinite. The situation only changed when Newton’s theory of gravitation was replaced by Einstein’s theory of general relativity.

2. Galactic and Extragalactic Astronomy

By 1900, the question of the location of the nebulae relative to the Milky Way system had moved to the forefront of what in retrospect can be termed observational cosmology. The crucial question was whether the nebulae, such as the Andromeda, were located within the Milky Way or whether they were large extragalactic structures apart from it and of a size and nature comparable to the Milky Way (Smith, 2019). According to the astronomer and astronomy writer Agnes Clerke (1890, p. 368), “No competent thinker, with the whole of the available evidence before him, can now, it is safe to say, maintain any single nebula to be a star system of coordinate rank with the Milky Way.” And yet there were astronomers, even competent thinkers, who defended the latter “island universe” hypothesis going back to a work of 1755 written by the famous philosopher Immanuel Kant. All the same, the majority probably shared Clerke’s view and its corollary that the Milky Way system was essentially identical to the material universe. Among the reasons for the uncertainty was not only lack of knowledge about the distances to the nebulae but also that the size of the Milky Way was a matter of dispute.

Due to works of Henrietta Leavitt, Ejnar Hertzsprung, and Harlow Shapley, by 1918, astronomers realized that if they could find just a single variable star of the type known as Cepheids in a nebula, it would be enough to determine its distance. All they had to do was to measure the luminosity period of the star. Unfortunately, no such Cepheid had been identified in either the Andromeda nebula or other nebulae.

During the same period in which the Cepheid method was developed by Shapley and others, Vesto Slipher at the Lowell Observatory in Arizona studied the spectra of the Andromeda and other spiral nebulae. Interpreting the measured shifts in wavelengths $∆\lambda$ as Doppler shifts implied that they were due to radial velocities v in accordance with Doppler’s formula $∆\lambda /\lambda =v/c$, where c denotes the speed of light. Slipher came to the surprising conclusion that most of the nebulae receded from the Earth with radial velocities up to 1,000 km/s or more (Way & Hunter, 2013). Although he refrained from interpreting the nebular redshifts cosmologically, his observations contributed to a revival of the island universe theory. After all, they suggested that the fast-moving spiral nebulae could not be gravitationally bound to the Milky Way. However, Slipher’s data constituted no proof, and in the absence of knowledge of the distances to the nebulae, the redshift phenomenon was circumstantial evidence only. By the late 1920s, nebular redshifts would become the observational basis for the expanding universe, but around 1920, no one thought along such lines.

The opposing views of the universe were discussed in public at a meeting in Washington, DC on April 26, 1920, subsequently known as the “Great Debate” (Smith, 1982). The discussants were Heber Curtis, who argued for the island universe, and Shapley, who argued against it and for an immensely large Milky Way. Shapley believed at the time that the galactic Milky Way system consisting of stars and nebulae had a diameter of no less than 300,000 light years and that this was incompatible with the island universe model. Although the Washington debate was enlightening, from a scientific point of view, it was disappointing insofar that it did not result in any consensus. No new observations were reported that unambiguously spoke for one of the rival theories and against the other.

For yet another few years, the situation remained a stalemate if now with increased support for the island universe theory from Slipher’s nebular redshifts and also from observations of novae in spiral nebulae. The stalemate ended drastically in the autumn of 1923, when Edwin Hubble, working at the Mount Wilson Observatory, made an important discovery. In the course of searching for novae in spiral nebulae, he noticed on a photographic plate of the Andromeda nebula a variable starlike object that he first thought was a nova. By examining the object more closely, he realized that it was not a nova but a Cepheid variable (Berendzen et al., 1984). The luminosity and period of the Cepheid told him that Andromeda was about 930,000 light years away from the Earth and thus could not possibly be a member of the Milky Way system even should it have the size that Shapley surmised.

Hubble’s celebrated discovery, which was only announced in early 1925, effectively settled the Great Debate in favor of the island universe theory, which was soon accepted by the astronomical community. The few opponents did not succeed in reversing the consensus view that the spiral nebulae were extragalactic worlds with a scale and structure comparable to that of the Milky Way. Now the spiral nebulae became extragalactic objects or just galaxies, and the universe at large came to be seen as a vast congregation of galaxies floating in a possibly infinite cosmic space. The advance that led to the new picture of the universe was observational in the classical astronomical tradition with no significant contribution from modern physical theory.

It is worth noting that the word Galaxy (with capital G) was traditionally a synonym of Milky Way, meaning that the term Milky Way Galaxy was redundant. Hubble never spoke of galaxies but consistently used nebulae or, to denote those outside the Milky Way, extra-galactic nebulae. His widely read semipopular book published in 1936 carried the title The Realm of the Nebulae. The presently used term galaxy was advocated by Shapley, but for a long time, it was unpopular and won broad acceptance only after Hubble’s death in 1953.

Armed with knowledge of the distances of more and more spiral galaxies (to use modern terminology), Hubble began in 1928 a new research project with the aim of relating the distances (r) to their redshifts $\left(∆\lambda /\lambda \right)$. There were theoretical reasons for expecting some kind of simple relationship, and on March 15, 1929, Hubble reported in the Proceedings of the National Academy of Sciences that the two quantities were nearly proportional (Nussbaumer & Bieri, 2009, pp. 114–120; Smith, 1982, pp. 180–186). He took most of the redshifts from Slipher’s work and, like other astronomers, translated them by means of Doppler’s formula into velocities (v). More precisely, he found that v = Hr with the constant of proportionality H (the Hubble constant or parameter) being around 500 km/s/Mpc. The symbol Mpc stands for megaparsec, a distance unit dating from 1913 and approximately equal to 3.3 million light years.

Hubble’s 1929 paper is famous and often identified with the discovery of the expanding universe, but this is to read too much in it (Hubble, 1929). Nowhere in his paper did he conclude that the galaxies are actually receding from us or otherwise suggest that the universe is in a state of expansion. The velocities appearing in the linear law were really redshifts that could be, but did not need to be, understood as real recessional velocities. Hubble cautiously referred to them as “apparent velocities.” For more than a year, his paper created but little attention, and it was far from hailed as a revolutionary contribution to cosmology, an observational proof that the universe is expanding. That only came 1 or 2 years later and then because of progress in theoretical cosmology.

3. Relativistic Cosmological Models

In February 1917, Albert Einstein read a paper to the Prussian Academy of Sciences in which he suggested a modification of the field equations of general relativity theory introduced 2 years earlier. The new equations became known as his “cosmological field equations” because they were aimed to describe the universe as a whole. According to Einstein, there was only one physically realistic solution to the equations corresponding to a particular cosmological model. This original Einstein model was characterized by being spatially closed and hence finite despite having no boundary. It was homogeneously filled with dilute matter and likewise could be ascribed a finite mass. On the other hand, Einstein assumed the universe to be temporally infinite, meaning that its radius of curvature did not vary in time.

To secure a static universe, which he and others thought was justified by current astronomical data, Einstein introduced in his equations a term that counteracted the gravitational attraction and was expressed by the average density of matter. The new term was given by an exceedingly small but nonzero “cosmological constant,” a new constant of nature similar to the attenuation factor that Seeliger had introduced in his Newtonian cosmology of the late 1890s. The cosmological constant is usually designated by the Greek letter Λ‎ (lambda). Einstein was not much concerned about the agreement of his model with astronomical observations. “Whether, from the standpoint of present astronomical knowledge, it is tenable, will not here be discussed,” he wrote at the end of his paper (Einstein, 1952, p. 188).

Later the same year, after having studied Einstein’s theory, the Dutch astronomer Willem de Sitter found another solution to the field equations that represented a very different world model. De Sitter’s universe was empty—that is, containing no matter at all—and yet it was spatially closed like Einstein’s. Moreover, de Sitter and his contemporaries conceived it as a model of a static universe and not, as later cosmologists would do, as representing an exponentially expanding universe. Also like Einstein’s model, the one proposed by de Sitter made use of the cosmological constant but in this case related in a different way to the radius of curvature. Of course, in de Sitter’s model, the cosmological constant was unrelated to the density of matter (which he took to be zero).

Although Einstein admitted that de Sitter’s alternative world model was mathematically correct, he insisted that it was wrong on physical grounds and that it disagreed with the spirit of general relativity theory. All the same, de Sitter’s alternative attracted much attention among the few physicists, astronomers, and mathematicians specializing in relativistic cosmology (Kerzberg, 1989). One reason why astronomers found the model to be attractive was that it, contrary to the Einstein world, predicted a kind of galactic redshifts that were possibly related to the redshifts observed by Slipher. In his 1929 paper on the linear redshift–distance law, Hubble suggested that the redshifts might in part be explained on the basis of de Sitter’s theory.

During the 1920s, physicists and astronomers discussed in detail which of the two rival relativistic theories of the universe was the most satisfactory. Among those active in the debate were the British astronomer Arthur Eddington, the German mathematician Hermann Weyl, and, in the United States, Howard Robertson and Richard Tolman. However, no agreement or clarification was reached. Only at the end of the decade did they vaguely recognize that there might be other models that incorporated features of both the Einstein world and the de Sitter world. With a few exceptions, they did not seriously contemplate that the universe as a whole might be dynamic rather than static. As late as 1929, Einstein wrote in an article for the Encyclopaedia Britannica,

Nothing certain is known of what the properties of the space-time continuum may be as a whole. Through the general theory of relativity, however, the view that the continuum is infinite in its time-like extent but finite in its space-like extent has gained in probability.

(Kerzberg, 1989, p. 335)

This view would not last for long.

The collapse of the static universe paradigm was foreshadowed by Alexander Friedmann, a Russian theoretical physicist, in a paper published in the widely read Zeitschrift für Physik in 1922. Contrary to other workers in the field, Friedmann concluded from a systematic investigation of Einstein’s field equations that there were other solutions than the two static models associated with the names of Einstein and de Sitter. The general solution, he proved, included a variety of dynamical world models, that is, models in which the radius of curvature depended on the time parameter. For example, he found a class of solutions that formally included expanding world models with an absolute beginning in time. Friedmann even referred to “the time since the creation of the world . . . to the present state” (Tropp et al., 1993, p. 158). Moreover, he derived from the field equations that there were solutions corresponding to a “cyclic universe” in which the size of the closed universe oscillated or pulsated between zero and a maximum value. And this was not all, for in a follow-up paper 2 years later, he discussed as the first one the possibility of hyperbolic and spatially infinite world models with a constant negative curvature.

Remarkably, although Einstein knew about Friedmann’s work and responded critically to it, it made no impact at all on mainstream cosmology. When the Belgian physicist and expert in general relativity Georges Lemaître in 1927 arrived at essentially the same equations as Friedmann, he was unaware of the work of his Russian colleague, who died prematurely in 1925. From a mathematical point of view, Lemaître’s paper, written in French and published in a not widely circulated journal (Annales Scientifique Bruxelles), was similar to Friedmann’s, but otherwise, it was strikingly different as it was of a physical rather than mathematical nature (Farrell, 2005; Lambert, 2011). Lemaître explicitly argued that the one and only real universe was expanding, and he supported the claim by explaining the galactic redshifts as a result of the expansion. As he emphasized, the redshifts were caused not by galaxies moving through space but by galaxies being carried with the expanding space. On this basis, he even derived the redshift–distance law announced by Hubble 2 years later and estimated for the “Hubble constant” a value of 625 km/s/Mpc.

Like Einstein’s static universe, Lemaître’s expanding version of it was positively curved and hence of finite size. By the late 1920s, Einstein had abandoned the idea of a cosmological constant, which he now considered a mistake, but in Lemaître’s thinking about the universe, it was a crucial and indispensable quantity that he kept to throughout his life. His work of 1927 is often said to be the beginning of big-bang cosmology, but this is a mistake. It pictured the universe as gradually evolving from the static Einstein state and thus without a definite beginning in time.

For about 3 years, Lemaître’s brilliant paper shared the fate of Friedmann’s work: It was ignored and most likely simply unknown to the majority of contemporary cosmologists. Remarkably, not a single scientific paper referred to it until the spring of 1930. Einstein was one of the very few who did know about the paper, but at the time, he was convinced that the universe was static and consequently rejected Lemaître’s model. Many years later, Lemaître recalled a conversation he had with Einstein in the autumn of 1927:

After some favorable technical remarks, he concluded by saying that from the physical point of view that appeared completely abominable to him. . . . I had the impression that Einstein was hardly aware of the astronomical facts.

(Lemaître, 1958, p. 131)

4. From Expanding Universe to Primeval Atom

Attitudes concerning the expanding universe changed dramatically in the early part of 1930 after Eddington belatedly studied or restudied Lemaître’s paper. He now understood that the expanding model answered most of the problems that had plagued cosmology and that it supplied Hubble’s empirical redshift–distance law with the necessary theoretical foundation. De Sitter reached the same conclusion, writing in a letter to Shapley in April 1930 that Lemaître’s theory of the expanding universe was “the true solution, or at least a possible solution, which must be somewhere near the truth” (Smith, 1982, p. 187). With the enthusiastic support of Eddington and de Sitter, the expanding universe became quickly accepted by most specialists, including Einstein—although in his case with some delay. The conversion from a static to an expanding world picture amounted to nothing less than a paradigm shift in cosmological thought.

This is not to say that everyone converted to the new paradigm, for some leading astronomers either adopted an agnostic attitude or suggested alternative models that explained the redshifts on the basis of a static universe. To the first group belonged Hubble, who never endorsed the expanding universe as uncontrovertibly true. The second group was represented by the innovative Swiss American astrophysicist Fritz Zwicky, who pioneered what became known as the “tired light” alternative widely discussed in the 1930s (Kragh, 2017). Nonetheless, by the early 1930s the expanding universe was broadly accepted and had even entered the public arena in the form of popular books such as de Sitter’s Kosmos from 1932 and Eddington’s best-selling The Expanding Universe from 1933.

The new revelation also gave rise to expanding universe models that differed from Lemaître’s original 1927 model or what, because of certain refinements added by Eddington, was often referred to as the “Eddington–Lemaître model.” Thus, in the spring of 1931, Einstein suggested a cyclic model similar to the one discussed by Friedmann 9 years earlier. The new Einstein universe belonged in a formal sense to the big-bang category insofar that it postulated a sudden beginning in time. The same was the case with the more important flat-space model with no cosmological constant that Einstein and de Sitter jointly proposed in 1932. According to this model, the universe expanded forever but at a continually slower rate. A simple calculation showed that its age t* was given by 2/3 times the inverse Hubble constant also known as the Hubble time (T = 1/H).

The influential Einstein–de Sitter model was of the big-bang type only in a formal mathematical sense and not in a physical–realist sense. In this respect, it differed markedly from Lemaître’s daring proposal in Nature of May 9, 1931, that the birth of this universe was due to a cataclysmic explosive event in the far past. The Belgian physicist imagined that there once existed what he metaphorically called a highly radioactive “primeval atom” in which the mass of the entire universe was concentrated. Only with the violent decay of this “atom” would space and time come into existence. The original super-atom was not an abstract pointlike “singularity” but a real body of finite size and density. Lemaître assumed, if only as an illustration or analogy, that this original body had a density of the same order as an atomic nucleus, namely, ${10}^{15}\mathrm{g}/{\mathrm{cm}}^3$. Otherwise, it had no physical properties and was therefore inaccessible to scientific inquiry (Kragh & Lambert, 2007).

Lemaître’s note in Nature was no more than an enigmatic and spirited speculation, but later in 1931, he developed it into a quantitative theory that deservedly has been called the first example ever of a relativistic big-bang universe. First of all, according to this model, the universe had only existed in a finite period of time. As a result of the original creation event, the universe had expanded rapidly, and due to the cosmological constant, it was presently in a state of accelerated expansion that would continue indefinitely. Why believe in this apparently fantastic scenario? Would there be any traces left of the original explosion that could be examined today? Lemaître thought so, arguing—incorrectly, it turned out—that the poorly understood cosmic rays were electrically charged fossils of the disintegration of the primeval atom several billion years ago. The cosmic rays, he said poetically, were “one of the most curious of the hieroglyphs of our astronomical library” (Kragh, 1996, p. 53).

Lemaître’s primeval atom theory received positive attention in the popular press but was met with reservation and sometimes hostility in the scientific community. A universe with literally a beginning in time—not only as a mathematical abstraction but as a physical reality—was widely considered unacceptable and especially so if the hypothesis was not supported by observational evidence. Although Eddington was one of the pioneers of the expanding universe, he flatly dismissed Lemaître’s new theory on both scientific and philosophical grounds. As he stated on one occasion, he found the notion of a created universe to be “repugnant.” Other astronomers and physicists likewise rejected the primeval atom theory or simply chose to ignore it. The Canadian astronomer John Plaskett said about Lemaître’s hypothesis that it was “the wildest speculation of all . . . an example of speculation run mad without a shred of evidence to support it” (Plaskett, 1932).

To make a long story short, through the 1930s, big-bang solutions to the cosmological field equations were rarely taken seriously or assigned physical reality. In the few cases where scientists entertained ideas somewhat similar to Lemaître’s, they did not refer to the Belgian physicist’s primeval atom hypothesis. To mention but one example, in 1938, the German physicist Carl Friedrich von Weizsäcker speculated that the early universe was extremely hot (about ${10}^{11}$ K) and of nuclear density. He thought that nuclear reactions in the primeval state of the universe might have produced not only some of the chemical elements but also the energy necessary for the subsequent expansion (Drischner, 2014). Although Weizsäcker’s cosmological scenario had elements in common with the earlier primeval atom hypothesis, he did not refer to Lemaître’s work. The idea of an original nuclear explosion or radioactive decay was only further developed after World War II and then in a version that owed little to Lemaître’s primeval atom theory of the early 1930s.

5. Two Rival Theories of the Universe

In the late 1940s, two completely different theories of the expanding universe saw the light of day. One of them, primarily due to the Russian American nuclear physicist George Gamow, was a typical big-bang theory insofar that it assumed that the universe was born in an extremely hot and dense state a finite time ago and since then had continued to evolve. The other theory of the universe, known as the “steady-state theory,” was proposed by three young Cambridge physicists, Hermann Bondi, Thomas Gold, and Fred Hoyle. It claimed that the universe was eternal in the past as well as in the future and that on a large scale, it had never changed significantly. Whereas Hoyle’s more mathematical version of the steady-state theory was based on equations similar to those proposed by Einstein, the Bondi–Gold version denied that the theory of general relativity was valid for the universe as a whole. Although the two versions differed in some respects, they led to the same picture of an everlasting and eternally expanding universe. The rivalry between the big-bang theory and the steady-state theory colored much of the cosmological debate from 1948 to about the mid-1960s (Kragh, 1996).

In a letter to Einstein of September 1946, Gamow stated the essence of his thoughts about the explosive beginning of the universe:

It is important to remember that in order to explain the present relative abundance of the chemical elements one must agree that in “the Days of Creation” the mean density and temp. of the Universe was ${10}^7\mathit{gm}/{\mathit{cm}}^3$ and ${10}^{10}$ K.

(Kragh, 1996, p. 110)

He first assumed that the mechanism of element formation was proton–neutron reactions, with the protons formed by the radioactive decay of a primeval hot soup of neutrons. To develop his theory quantitatively, Gamow entered a close and fruitful collaboration with two assistants, Ralph Alpher and Robert Herman. The three physicists realized that the high temperature required the very early universe to be composed mainly by electromagnetic radiation and not by matter. As the universe expanded and grew colder, the ratio between radiation density and matter density would decrease and eventually lead to the present matter-dominated universe.

Importantly, in a brief paper of 1948, Alpher and Herman argued that the original radiation had by now shifted to the microwave area and with a very low intensity corresponding to a temperature of around 5 K. In other words, they predicted the existence of the weak cosmic microwave background that many years later would be detected and then be interpreted as convincing evidence for the hot big-bang theory. However, at the time, the prediction failed to attract attention among physicists and astronomers outside Gamow’s small research group (Alpher & Herman, 2001).

The calculations of Gamow and his collaborators resulted in promising results for the cosmological formation of helium but not for the heavier elements. All attempts to build up elements heavier than helium by means of realistic thermonuclear reactions failed, which was widely seen as a grave difficulty for Gamow’s theory of the early universe. This theory reached its climax in 1953, when Alpher, Herman, and James Follin presented a detailed and much improved version of it. The complex calculations of the three authors followed the evolution of the early universe from t =${10}^{-4}$ seconds after the initial explosion at t = 0 to around 10 minutes later. In this brief period of time, they found that helium would be formed in an amount of 32% of the total mass of the universe. This would turn out to be an approximately correct figure, but in the early 1950s, there were no observational results with which the theoretical figure could be compared. For this reason, it was ineffective as an argument in favor of the big-bang theory.

Strangely, from a modern perspective, with the work of Alpher, Herman, and Follin, big-bang cosmology came to an almost complete halt and was only resurrected a decade later by other physicists. From 1954 to 1963, only a single scientific paper was devoted to the theory of Gamow and his collaborators.

There were additional reasons why the majority of physicists and astronomers tended to disbelieve finite-age relativistic cosmologies in general and the big-bang theory in particular. Most of the evolutionary models belonging to this class predicted an age of the universe t* smaller than the Hubble time and therefore smaller than the age of the stars and even of the Earth! Based on Hubble’s observations, the accepted Hubble time was 1.8 billion years, whereas the Earth was known to have an age of 4 billion years or possibly more (Kragh, 1996, pp. 73–79, 271–273). The “age paradox” only became less paradoxical when astronomical measurements in the mid-1950s showed that the value of the Hubble time was too low by a factor of 2 or more. By the end of the decade, a value of 10–15 billion years was broadly accepted (the present values are T = 14.4 billion years and t* = 13.7 billion years).

Of course, since there was no beginning of the universe posited by the steady-state theory, there was also no age paradox. In its early version, this theory built conceptually on the “perfect cosmological principle,” according to which the universe in its large-scale features was homogeneous in both space and time. It followed from the principle that the average density of matter must remain constant and therefore that new matter, perhaps in the form of hydrogen atoms, must continually be created as the universe expands. Although the necessary creation rate was miniscule, no more than ${10}^{-43}\mathrm{g}/\mathrm{s}/{\mathrm{cm}}^3$, it seemed to violate the fundamental law of energy conservation and was for this reason a most controversial element in the new cosmology.

Contrary to the wide class of relativistic evolution theories, the theory of Hoyle and his allies led to several definite predictions that could be tested by means of experiment and observation not only in principle but also in practice. In philosophical terminology, the theory was highly falsifiable. For example, the steady-state universe was infinite and geometrically flat, and it expanded exponentially at a constant rate. The predicted matter density happened to be the same as that of the Einstein–de Sitter model—namely, around $5\times {10}^{-28}\mathrm{g}/{\mathrm{cm}}^3$. Moreover, as shown by the Irish physicist William McCrea in 1950, it followed from the steady-state theory that there was a definite age distribution of the galaxies and that their average age was equal to one third times the Hubble time. McCrea was during the 1950s an active and valuable supporter of the steady-state theory, which he preferred for both scientific and philosophical reasons.

There was no hot and dense state in the steady-state theory, which consequently had to account for the abundance and distribution of the elements solely in terms of nuclear reactions taking place in stellar bodies of various kinds. In an important collaborative work with William Fowler, Margaret Burbidge, and Geoffrey Burbidge, Hoyle produced in 1957 a comprehensive theory of stellar nucleosynthesis that explained the formation of practically all elements and isotopes from the lightest to the heaviest. This pioneering theory is generally known under the acronym B2FH, a reference to Burbidge, Burbidge, Fowler, and Hoyle. The successful B2FH theory was not explicitly derived from steady-state assumptions, but because it made no use of the hypothesis of a big bang, it was sometimes taken as indirect support of steady-state cosmology.

6. A Cosmological Controversy

The more than decade-long discussion concerning the foundation of cosmology and the evolution of the universe was in part of a philosophical nature and attracted no less attention among philosophers than among physicists and astronomers (Kragh, 2022). By the mid-1950s, it was still uncertain if the universe as a whole could be the subject of scientific inquiry in the same way as parts of it such as stars and galaxies. If cosmology were accepted as physical science, what would the criteria for theory choice be?

Characteristic for the climate at the time, in 1954, Bondi discussed with Gerald Whitrow, an esteemed physicist and astronomer, the question “Is Physically Cosmology a Science?” in the pages of the British Journal for the Philosophy of Science. The question was seriously meant, not rhetorical. Inspired by the influential Austrian British philosopher Karl Popper, Bondi emphasized falsifiability as the hallmark of science, and from this perspective, he saw no reason why cosmology should not be a genuine science. Whitrow was not so sure. Bondi also suggested that since the steady-state theory was undoubtedly more easily falsifiable than the evolution theories based on general relativity, from a methodological point of view, the first theory was a better choice than the latter.

Despite the role played by methodological and other philosophical considerations in the cosmological controversy, it was first and foremost about observations and how they related to the predictions of the two theories. It was generally agreed that ultimately, the controversy would be settled by new observational and experimental data.

If the curvature of space and the manner in which the rate of expansion increased or decreased could be determined, it would be possible to discriminate between the two rival cosmological models. This approach was followed by the leading American astronomer Allan Sandage, who in 1956 reported observations that apparently disagreed with the steady-state theory. However, by closer inspection, his data turned out to be inconclusive, and neither did later results from optical astronomy unambiguously refute the theory defended by Hoyle, Gold, and Bondi. The new science of radio astronomy promised to do better. Radio astronomy turned into radio cosmology in 1955, when Martin Ryle and his group at Cambridge University concluded that the distribution of radio sources or what at the time was called “radio stars” disagreed with the steady-state theory (Sullivan, 1990).

However, the announcement that the theory had now been disproved was premature. The Cambridge results were contradicted by measurements made by Bernard Mills and his team of radio astronomers in Sydney who obtained quite different results from the Southern Hemisphere. For a while, it seemed that radio astronomy was as unable as optical astronomy to distinguish between the two competing models of the universe. This was not the case, though, for in the early 1960s, new and more reliable data from both Cambridge and Sydney showed otherwise—namely, that observations of radio sources provided strong evidence against the steady-state theory. The consensus among the radio astronomers did not conclusively refute the theory, but it left it in a seriously weakened state. Hoyle and what remained of the steady-state camp admitted as much, and yet they thought that the discrepancy between observations and theory was not inevitable. Perhaps it would disappear in a suitably revised version of the steady-state theory.

Although the half-forgotten Alpher–Herman prediction of a cosmic background radiation played almost no role in the cosmological controversy, in the early 1960s, a few physicists came to the same conclusion. In about 1963, the Princeton physicist Robert Dicke hypothesized a cosmic black-body radiation originating from the collapse of a pre–big-bang universe and its subsequent rebirth. Dicke’s speculation about a cyclic universe was unfruitful except that it revived the idea of a cosmic background made up of low-intensity electromagnetic waves. In 1964, James Peebles, also at Princeton, calculated the properties of the assumed radiation, which he estimated to have a present temperature of around 10 K. Neither Dicke nor Peebles were at the time aware of the old prediction of Alpher and Herman. Nor were they aware of the results obtained by two researchers at Bell Laboratories while examining an antenna for use in radio astronomy (Peebles et al., 2009).

Arno Penzias and Robert Wilson found to their surprise an excess temperature in their antenna of 3.5 K, which somehow was of cosmic origin but without associating their finding with cosmological models of the early universe. When Dicke and his group in Princeton were informed about the anomalous excess temperature, they immediately understood that Penzias and Wilson serendipitously had discovered a fossil radiation from the very early universe. The discovery was made public in the New York Times of May 21, 1965 (“Signals Imply a ‘Big Bang’ Universe”), and the details were given in the July 1965 issue of Physical Review, which included the observational results of Penzias and Wilson alongside with the big-bang interpretation due to Dicke and his collaborators.

When further work confirmed the background radiation at other wavelengths than the one used by Penzias and Wilson (7.3 cm), there was little doubt: It was the result of a very early and hot phase in the history of the universe and therefore crucial evidence that the steady-state theory was wrong. Hoyle and his collaborator Jayant Narlikar disagreed, but their sustained attempts to explain the data in terms of starlight and in agreement with a modified steady-state theory won no acceptance.

The discovery of the cosmic microwave background had important implications for the nuclear processes leading to the formation of helium in the early universe. Calculations of this kind had been performed by Gamow’s group in the early 1950s, and they now reappeared in a much-improved version. In 1966, Peebles derived from big-bang calculations a cosmic abundance of helium of 26%–28% by mass depending on the value of the present density of matter. The figure agreed nicely with earlier observations by Donald Osterbrock and John Rogerson, who found around 25%. The amount of helium in the universe thus provided one more confirmation of the hot big-bang model.

Yet another confirmation, should one be needed, was provided by the strange quasars discovered in 1963. Recognized to be extragalactic objects, the quasars were cosmologically important in the same way that the spiral galaxies were. In a study of 1966, Dennis Sciama and Martin Rees investigated the distribution of quasars and their redshifts in relation to the prediction of the steady-state theory. They concluded that observational data and theory could not possibly be brought into agreement. Sciama had for years defended the steady-state theory, but now he lost his faith.

7. A New Cosmos

The presently accepted theory of the universe and its evolution in time is conveniently labeled the “big-bang theory,” but this is a slightly anachronistic phrase if used for the early development of the theory. The term big bang was originally coined by Hoyle, an archenemy of this kind of cosmological theory, in a BBC broadcast of March 20, 1945. “I cannot see any good reason for preferring the big bang idea,” he said. “Indeed it seems to me to be in the philosophical sense a distinctly unsatisfactory notion, since it puts the basic assumption out of sight where it can never be challenged by a direct appeal to observation” (Kragh, 2014). One might expect that the catchy phrase quickly caught on and was frequently used in the cosmological controversy, but this was far from the case. Gamow disliked the term, and although Hoyle referred to it in his popular book The Nature of the Universe from 1950, he only used it again in 1965. In their seminal paper in the Astrophysical Journal of July 1965, Dicke and his coauthors did not employ the term but instead wrote about the “primeval fireball” as the state of the early universe giving birth to the microwave background. In fact, the later so popular phrase only appeared insignificantly in the scientific literature until the early 1970s, when it finally gained momentum.

Nonetheless, the year 1965 can reasonably be seen as the beginning of a new chapter in the annals of cosmology, the year in which the modern big-bang theory was born. Of course, the new theory was not accepted instantly or by all physicists and astronomers (not to mention the philosophers). For a decade or more, it was resisted by a substantial minority of scientists who suggested alternative cosmologies without an explosive beginning of the universe at large. By the late 1960s, former steady-state supporters such as Bondi, Gold, Sciama, and McCrea had either abandoned the theory of an eternal universe or withdrawn from research in cosmology. However, Hoyle continued until his death in 2001 to fight what he thought was the semireligious and unacceptable big-bang orthodoxy. Together with Narlikar and a few other collaborators, he developed a series of alternative theories that had in common with the original theory, only that they preserved the idea of an eternal universe with continual creation of matter.

Another kind of alternative cosmological theory was promoted by the prominent Swedish physicist Hannes Alfvén, who in the 1960s introduced a “plasma cosmology” based on the assumption of equal amounts of matter and antimatter in the universe. Alfvén’s favored cosmology shared with Hoyle’s that it denied a beginning of the universe as a whole, but otherwise it was quite different. None of the two alternatives succeeded in attracting wide interest, and by the 1980s, they had become marginalized. The same was the case with versions of the old “tired-light” hypothesis, which not only denied the big bang but also that the universe was expanding. By the late 1980s, respected journals of astrophysics and cosmology no longer accepted papers belonging to the tired-light category.

Less than a decade after the discovery of the cosmic microwave background, the hot big-bang theory of the universe had acquired an almost paradigmatic status. At the same time, cosmology experienced a strong quantitative growth such as indicated by bibliometric data (Ryan & Shepley, 1976). The annual number of research articles on cosmology had on average been around 30 in the period from 1950 to 1962, a very low number. Between 1962 and 1972, it increased from 50 to 250. Another indication of cosmology’s growing maturity as a scientific discipline is the emergence of textbooks on the subject and the content and style of those books. Before 1970, there were only a couple of textbooks, and the most used of these, Bondi’s Cosmology from 1952 with a revised edition of 1956, covered very different and incompatible theories of the universe. Peebles’s Physical Cosmology from 1971 was entirely different as it essentially identified cosmology with the new theory of an explosive beginning of the universe and its subsequent evolution governed by the laws of general relativity. Peebles’s book was quickly followed by others in the same genre, of which Steven Weinberg’s Gravitation and Cosmology from 1972 was arguably the most important.

For the first time ever, cosmology became a full-time professional occupation for trained scientists with a shared view of what their field of research was all about. With the professionalization and increased academic status followed a de facto exclusion of the philosophers and amateur scientists who had previously been significant actors in the development of cosmological thoughts. The designation “cosmologist” had traditionally been reserved for philosophers speculating about the meaning of the universe and not widely used, if used at all, by the scientists engaged in cosmological research. Now some astronomers and physicists slowly began to perceive themselves as cosmologists in a different and strictly scientific sense.

It is tempting to speak of the conceptual change that occurred in cosmology in the 1960s as a “revolution” and the victorious big-bang theory as a new “paradigm.” However, the terminology is hardly appropriate if understood in the strong sense of the terms associated with Thomas Kuhn’s philosophy of science (Marx & Bornmann, 2010). Something new and very important did happen in the period, and yet the new picture of the finite-age universe was not incommensurable with earlier views such as held by Lemaître and Gamow. After all, as far as fundamental physics is concerned, the new consensus theory rested safely on old ground, namely, general relativity and nuclear physics as described by quantum mechanics. Whereas the change was not truly revolutionary in Kuhn’s sense, in a sociological perspective, it led to a kind of cosmological paradigm that, despite all later developments, has lasted until the present.

When James Peebles received the Nobel Prize in physics in 2019, the universe looked significantly different from the one he had co-created half a century earlier (Peebles, 2020). The modern developments are not parts of the present article, but three of the most important discoveries deserve brief mention:

8. Conclusion

Although cosmology as a physical science only emerged in the 20th century, the quest of understanding the universe has always been a challenge for philosophically minded people. The kinds of “cosmologies” discussed by Anaximander and other pre-Socratic philosophers—or those discussed in the Middle Ages by Thomas Aquinas and other scholastics—were of course very different from the cosmological models based on modern physical theory. And yet it would be a mistake to believe that people’s present ideas of the universe have nothing at all in common with the speculations of the past. Across a time span of more than two millennia, there has been a remarkable measure of permanence in the kind of cosmic questions people have wanted to know the answers to.

While some of the old questions have now been deemed scientifically meaningless and excluded from the scientific discourse about the universe, others are still part of what cosmologists are concerned with either directly or indirectly. To the first category belongs the question of the purpose of the universe and God’s role in the cosmic history. But there are other and equally majestic questions that may well be answered by accepted scientific methods and therefore belong, at least in principle, to the domain of modern cosmology. Is the world at large finite or infinite? Has the world always existed, or did it once come into existence? If there were an absolute origin, how can it possibly be explained without recourse to a divine creator? To mention but one more of the old but not obsolete questions: Does advanced life exist elsewhere in the universe?

The latter question is not only of astronomical relevance because of the discovery since the 1990s of numerous exoplanets revolving around stars different from the Sun. It also relates to cosmological thinking in the form of the so-called anthropic principle introduced by the astrophysicist Brandon Carter in 1974. According to the simplest version of this principle or argument, carbon-based life could not have originated in a universe evolving just slightly different from the one observed . In another version, it states that important aspects of the universe can only be explained by the existence of human life. Whatever the precise meaning of the controversial anthropic principle, it has attracted massive attention among cosmologists as well as philosophers (Carr, 2007; Vidal, 2014). While some see it as a new foundation of cosmological thinking, others deny its scientific legitimacy. Interestingly, the present debate concerning the anthropic principle and its consequences is in some ways strikingly similar to the cosmological controversy of the past between the steady-state theory and relativistic evolution theories.