Who discovered the expansion of the Universe?

 

 

Søren Larsen

Department of Astrophysics / IMAPP, Radboud University, Nijmegen

s.larsen@astro.ru.nl

 

 

The discovery that we live in an expanding Universe was arguably one of the most fundamental scientific breakthroughs of the 20th century. The popular version of the story, as it is often told, goes something like this: Upon having developed his theory of General Relativity (GR) in 1915, Albert Einstein found that the solutions to the field equations did not allow for a static Universe, but instead required the Universe to be contracting or expanding. As this seemed unsatisfactory to Einstein, he introduced an extra term involving the cosmological constant, Λ, in the field equations. The effect of Λ is to counteract gravity, thereby allowing for a static Universe. Over a decade later, in 1929, Edwin Hubble showed that the radial velocities of galaxies are proportional to their distances. The Universe indeed appeared to be expanding, and the relation between distance and velocity (or redshift) has since been known as “Hubble's law”. The original prediction of a non-static Universe thus appeared to be correct after all, and Einstein renounced the introduction of the cosmological constant as his “greatest blunder”. Recently, there has been a growing awareness of the work by Georges Lemaître, and it has been proposed to rename the velocity-distance relation to the “Hubble-Lemaître law”.

The real story is more complicated and involves many other individuals who made substantial contributions throughout the 1920s, with Hubble entering the stage relatively late. Indeed, the story starts even before Einstein had finalised GR, at a time when nobody was concerned with the dynamics of the Universe at large. In the following I attempt to summarise some of the main milestones in roughly chronological order. I will focus on the period up to about 1930, after which the existence of the velocity-distance relation and its interpretation as evidence of an expanding Universe was no longer in much doubt. I hope that I have not overlooked any major contributions, but would welcome any comments on this text.

 

Radial velocity measurements of spiral nebulae

 

A reasonable place to start is 1913. In this year, Vesto Slipher at Lowell Observatory published a short paper in which he presented his spectroscopic observations of the Andromeda nebula, M31,  which he had carried out in the fall of 1912 [1]. (Non-stellar diffuse objects were then generally referred to as nebulae. I will use the term “nebulae” and the modern term “galaxies” interchangeably.) Slipher's aim was to measure the radial velocity of the nebula via the Doppler shift of its spectral lines, a technique which had by then been successfully applied to individual stars already for a few decades. It was known that stars typically have Doppler shifts indicating positive or negative radial velocities of up to a few tens of km/s, and there were hints that the dispersion depended on spectral class.  However, obtaining good spectra for non-stellar diffuse objects was a much more challenging enterprise.

Slipher was not the first to obtain a spectrum of M31, which was known to exhibit absorption lines similar to those observed in the spectrum of the Sun [2], but nobody had so far measured the radial velocity of any galaxy. Using a spectrograph optimised for this purpose and exposures lasting up to 3 nights, Slipher was able to record several spectra of sufficient quality to measure a radial velocity of about -300 km/s for M31, entirely consistent with modern measurements.  He noted that the magnitude of the velocity was the greatest observed so far, and that this “raises the question whether the velocity-like displacement might not be due to some other cause, but I believe we have at present no other interpretation for it.”. He further remarked that “extension of the work to other objects promises results of fundamental importance, but the faintness of the spectra makes the work heavy and the accumulation of results slow”.

By 1915, Slipher had collected spectra of 15 spiral nebulae [3]. There were two surprises: first, as was true for M31, the velocities were all much greater in an absolute sense than those typical of stars. Second, M31 turned out to be atypical in having a negative radial velocity; nearly all other spiral nebulae had positive radial velocities of up to several hundred km/s. Slipher had no real explanation for these results, but suggested that there might be a north-south asymmetry with larger positive velocities on average for nebulae on the southern side of the Milky Way (M31, M32, and M33 constituted half of the northern sample), and that edge-on nebulae appeared to have higher velocities. In hindsight, these trends were mostly spurious and a result of the small sample size. What was clearly significant, however, was Slipher's observation that “the average velocity of the spirals is about 25 times the average stellar velocity”.  At this point, two years before Einstein's paper on the cosmological implications of GR, Slipher had no way of anticipating the cosmological significance of his measurements. He briefly discussed the velocities of the nebulae in the context of contemporary work on individual stars, noting that “This great velocity would place these nebulae a long way along the evolutional chain if we undertook to apply the Campbell-Kapteyn discovery of the increase in stellar velocity with “advance” in stellar spectral type”.

It is worth pausing to reflect on the state of knowledge in 1915. Not only did the notion of an expanding Universe still lie in the future, but the very concept of the Universe as we understand it today had yet to emerge. While the hypothesis of “island universes” dates back to the 18th century, the nature of spiral nebulae and their relation with the Milky Way was still largely unknown in 1915 - there wasn't a clear concept of Galactic versus extra-Galactic astronomy, let alone cosmology in the modern sense.

Apparently Slipher's radial velocity measurements were not completely uncontroversial. In 1917, he felt compelled to publish a note responding to a suggestion by Reynolds that the velocities might not be reliable [4]. Slipher carried on with his work and kept finding increasingly large velocities (up to 1800 km/s in 1918 for NGC 584, which would hold the record for more than a decade). He also observed other types of objects, and found that globular clusters did not share the same large radial velocities as the spiral nebulae, but were instead intermediate between individual stars and the spirals in this respect.

 

Early cosmological models and curved space-time

 

Einstein's 1917 paper, in which the cosmological constant was introduced, did not mention Slipher's radial velocity measurements or any other specific astronomical observations, apart from general remarks about stellar velocities being small compared to the speed of light [5]. In the same year, W. de Sitter published an alternative solution to the field equations, in which the density of matter in the Universe was assumed to be negligible. In modern terminology, de Sitter's model has ΩΛ = 1 and Ωmatter = 0, which leads to an exponentially expanding Universe, although the coordinates chosen by de Sitter made this point somewhat obscure. Nevertheless, de Sitter realised that his model predicted a relation between distance and redshift. He wrote “... the frequency of light-vibrations diminishes with increasing distance from the origin of co-ordinates. The lines in the spectra of very distant stars or nebulae must therefore be systematically displaced towards the red, giving rise to a spurious radial velocity.” [6].

One of the major novel concepts in GR was, of course, the notion of curved space-time, and it was an intriguing idea that it might be possible to determine the radius of curvature from astronomical observations. In de Sitter's model, with his choice of coordinates, both space and time were curved and the radius of curvature Rc was related to the redshift z and distance D. Although World War I made communication more difficult, de Sitter was evidently aware of Slipher's measurements, and he listed radial velocities for three spirals for which Slipher's velocities had been confirmed by other observers (Pease at Mount Wilson, Moore at Lick). In addition to M31, these were NGC 1068 (+925 km/s) and NGC 4594 (+1185 km/s). However, distances to spiral nebulae were still unknown so de Sitter could only guess. His guess was 100 kpc, from which he derived a radius of Rc=3×1011 AU (about 1.5 Mpc).  While he admitted that “this result, derived from only three nebulae, has practically no value”, determining the radius of curvature would be a recurrent theme in several studies over the coming years.

A brief intermezzo about distances: it is often stated or implied that the problem of the distances to spiral nebulae was first solved with Hubble's discovery of Cepheid variables in M31 in 1923 (published in 1925). However, it is worth noting that Knut Lundmark, while working on his PhD thesis in Uppsala, estimated the distance to M31 to be about 200 kpc already in 1919, using observations of novae [7]. While Lundmark's distance to M31 was about a factor of four too small compared to modern measurements, it was similar to the distance derived by Hubble from Cepheids several years later. Lundmark's paper appears to have had relatively little impact at the time, perhaps because it also included a wide range of distances estimated by other methods (some of which were much further from the correct value). We shall return to Lundmark shortly.

The now familiar (locally) linear relation between redshift and distance was first given explicitly by H. Weyl in 1923 for de Sitter's model [8]. He expressed the relation as Δλ/λ = tan(D/R) for wavelength λ, distance D and radius of curvature Rc. In the limit of D << Rc, this is essentially “Hubble's law”, v = H0 D, but expressed in terms of the curvature radius instead of what we would today call H0. Like de Sitter, Weyl was clearly aware of Slipher's measurements, remarking: “Concerning the relation of our result to experience, namely the strong redshifts of the spectral lines of spiral galaxies found by astronomers, I refer to Eddington in his new book.”. Eddington's book from 1923 [9] contains a chapter in which Einstein's and de Sitter's models are discussed in detail, and also includes a table with 41 radial velocity measurements obtained by Slipher. In the book, Eddington cautiously remarked that “The great preponderance of positive (receding) velocities is very striking; but the lack of observations of southern nebulae is unfortunate, and forbids a final conclusion.”, but he proceeded to discuss the interpretation of the positive radial velocities in the context of de Sitter's model. Essentially the same formula for the velocity-distance relation was given by L. Silberstein in 1924 [10]. Silberstein's paper is interesting in that it attempts to use the velocity-distance relation as a distance indicator, albeit on somewhat shaky grounds: Silberstein attempted to derive the curvature radius from observations of globular clusters, and then applied it to estimate the distances to spiral nebulae using Slipher's radial velocity measurements. This is a reminder that observational cosmology was still very much in its infancy, and it was not yet clear which objects were the best cosmological probes.

 

Observational tests: finding the radius of curvature

 

Already in the early 1920s, astronomers were starting to look for correlations between Slipher's radial velocity measurements and various properties of the galaxies. The predominantly positive radial velocities had caught the attention of C. Wirtz in Kiel already in 1918 [11]. Wirtz commented that it appeared as if the system of spiral nebulae was drifting away from the solar system with a mean velocity of 656 km/s, but also pointed out that this interpretation relied on the assumption that the shifts of the spectral lines were caused by the Doppler effect. In 1922, Wirtz used a compilation of 29 radial velocity measurements available by then to show that the radial velocities tended to increase with decreasing apparent brightness of the galaxies (increasing apparent magnitude) [12]. By 1924, 42 radial velocities were available, and Wirtz found a strong inverse correlation between the apparent diameters of the spirals and their radial velocities. He wrote “Er bleibt also so kein Zweifel, daß die positive Radialbewegung der Spiralnebel mit zunehmender Entfernung ganz erheblich anwächst” – “There is no doubt that the positive radial velocities of the spiral nebulae increase considerably with increasing distance” [13]. Wirtz discussed these results in the context of de Sitter's model (with a reference to Eddington's book), but did not appear to be aware of the prediction that the redshift-distance relation should be linear. Instead, he gave the relation as v = 2200 - 1200 log Dm for velocity v in km/s and apparent diameter Dm in arcminutes, implying a logarithmic dependence of the velocity on distance. In 1936, Wirtz published a brief note reminding his peers of these results [14]. Plots of his data are given by Seitter & Duerbeck [15].

Knut Lundmark's papers from 1924 and 1925 addressed a wide range of issues in the nascent field of extragalactic astronomy [16][17]. In the 1924 paper, entitled “The determination of the curvature of space-time in de Sitter's world”, Lundmark plotted the radial velocities of the spiral nebulae (“mainly due to the wonderful spectrographic work performed at the Lowell Observatory by Dr. V. M. Slipher”) versus their distances relative to that of M31 (based on apparent diameters and magnitudes). The plot showed “that there may be a relation between these two quantities, although not a very definite one.” Lundmark reiterated here his best estimate of the distance to M31 as 200 kpc, but suggested that the correct value might turn out to be closer to 500 kpc. He also investigated velocity-distance relations for other types of objects, but commented that “Dr. Silberstein has not given, and will probably not be able to give, any justification for the use of the velocities of the globular clusters for a determination of R”. In the 1925 paper, Lundmark fitted the velocity-distance relation with a quadratic function, obtaining v = 513 + 10.365 D - 0.047 D2 for D in units of the distance to M31, and remarked about the coefficient m=-0.047 that “It is thought that m, although inaccurately known, still expresses a real phenomenon.”. If this were indeed true, the radial velocities would reach a maximum of 2250 km/s at 110 times the distance of M31.

G. Strömberg carried out an analysis similar to Lundmark's, and concluded that “the correlation-coefficients between distance and radial velocities give no clear evidence of curvature in either De Sitter's or Silberstein's sense[18].

 

Beyond de Sitter's model

 

While de Sitter's model could account for the positive redshifts of the spiral nebulae, it remained unsatisfactory in that galaxies (and other objects) were treated merely as test particles in an otherwise empty Universe. Einstein's model, on the other hand, contained matter, but was static. Eddington commented that “It seems natural to regard de Sitter's and Einstein's forms as two limiting cases, the circumstances of the actual world being intermediate between them.”. Yet, it was not immediately obvious how to find intermediate models.

That Einstein's and de Sitter's models were special cases of a more general set of non-static solutions of the field equations had, in fact, been demonstrated by Alexander Friedman in 1922 [19]. Friedman derived the equations governing the dynamics of the Universe that now bear his name, and found that the solutions included expanding and oscillating Universes that contained matter. He also noted that the expanding solution implied a finite age of the Universe. In 1924 he extended the analysis to cases of negative curvature. For several years, however, Friedman's papers went largely unnoticed, and they did not explicitly discuss the distance-velocity relation. Indeed, the 1922 paper ends by concluding that our knowledge is insufficient to determine which of the models is the correct one. It is well known that Einstein did not particularly like Friedman's expanding models [20][21].

In a paper published in 1925, G. Lemaître discussed various strange properties of de Sitter's solution, and showed that it could be recast in the now familiar interpretation of an exponentially expanding Universe with no curvature [22]. He (re-)derived the locally linear distance-redshift relation and wrote: “Our treatment evidences this non-statical character of de Sitter's world which gives a possible interpretation of the mean receding motion of spiral nebulae”. However, he concluded that “De Sitter's solution has to be abandoned, not because it is non-static, but because it does not give a finite space without introducing an impossible boundary”. The same year, Lemaître presented the paper at a meeting of the American Philosophical Society, and also arranged to talk to Hubble about extragalactic distances [23]. Later, in 1927, Lemaître published a paper in which he showed that Einstein's and de Sitter's models were special cases of a more general set of solutions (apparently unaware of Friedman's earlier work) and that a linear distance-redshift relation was expected not only for the de Sitter model [24]. In the same paper, Lemaître combined 43 radial velocities (mostly from Slipher, as usual) with estimates of the distances to provide an estimate of the slope of the velocity-distance relation, about 625 km/s/Mpc. The distances were based on Hubble's estimate of the mean absolute magnitude of the galaxies, M = -15.2, published the previous year [25], so it is not surprising that Lemaître’s expansion coefficient was very similar to that determined by Hubble a couple of years later.

In his 1927 paper, Lemaître did mention the work of Lundmark and Strömberg, noting the large scatter in their velocity-distance relations, which had its origin in the still very uncertain distances.  There is no indication that Lemaître found a significantly smaller scatter, although the use of Hubble's estimate of the absolute magnitudes allowed him to put the distances on an absolute scale with more confidence (though not necessarily more accuracy) than others before him. He did not claim that the data actually demonstrated a linear velocity-distance relation, but simply divided the average velocity by the average distance of the galaxies in the sample. In the 1931 English translation of the paper [26], which was published in the Monthly Notices at the request of Eddington, Lemaître omitted the paragraph in which the numerical value of the expansion coefficient was discussed, noting in his letter to the editor that “I did not find advisable to reprint the provisional discussion of radial velocities which is clearly of no actual interest.” [27].

Lemaître‘s paper provided an important demonstration that de Sitter's model is not unique in predicting a distance-redshift relation. Still, Lemaître‘s preferred model differed from our modern cosmological picture. In the model considered in the 1927 paper, the Universe was curved and started with a finite radius in the infinite past. Lemaître estimated the initial radius to be about 20 times less than the current value, so that the largest observable redshift would be z=20.  He had thus found a solution to the problem of infinite radius, which had bothered him in the de Sitter model, and the solution allowed for an expanding Universe containing matter.

H. P. Robertson was addressing the difficulties with de Sitter's solution around the same time as Lemaître. In a paper published in 1928, Robertson showed that de Sitter's model was naturally interpreted as an exponentially expanding Universe [28], apparently independently of Lemaître’s 1925 paper. Like Lemaître, Robertson also showed that the resulting space-time was flat, but he further realised that the volume accessible to observations would nevertheless be contained within a finite horizon R which would be related to the expansion rate. Robertson wrote: “Comparing the data given by Hubble concerning the value of l [distance] for the spiral nebulae with that of Slipher concerning the corresponding radial velocities, we arrive at a rough verification of (17) [the linear distance-velocity relation] and a value R=2×1027 cm.”. This corresponds to an equivalent Hubble constant of 460 km/s/Mpc. Robertson referenced Hubble's 1926 paper for the distances and Eddington's book for Slipher's velocities, and although it is not clear how exactly these data were combined to determine the value for R, or to what extent they supported the idea of a linear distance-velocity relation, it is thus not surprising that Robertson found an expansion coefficient similar to that obtained by Lemaître the previous year, and by Hubble a year later. In 1929, Robertson showed that Einstein's and de Sitter's models were special cases and again discussed the redshifting of the light from distant sources [29]. The paper includes references to Friedman and also contains a note (in the context of de Sitter's solution) that “I have since discovered that these coordinates have also been employed by G. Lemaitre, Jour. of Math. and Phys., 4, 188 (1925), and wish to take this opportunity to correct the omission of reference to this work in my previous paper.”. However, Robertson was at this point apparently not aware of Lemaître’s 1927 paper.

 

Hubble and the velocity-distance relation

 

At last, this brings us to Hubble's work. Before discussing the 1929 paper on the distance-velocity relation, it is interesting to consider Hubble's 1926 paper on Extra-galactic Nebulae [25], in which he introduced his classification scheme and discussed a variety of properties of the nebulae - morphology, absolute and apparent magnitudes, sizes, spatial distribution, etc. - in considerable detail. At the end of the paper, Hubble estimated the radius of curvature using the mean density obtained from the galaxy counts and assuming Einstein's static model. Radial velocities were only mentioned in a brief remark that mainly served to argue that M31 and M32 are physically associated: “M32 is generally assumed to be associated with the great spiral M31, because the radial velocities are nearly equal and are unique in that they are the only large negative velocities that have been found among the extra-galactic nebulae”. There is little hint that the velocity-distance relation was on Hubble's mind at this point.

This had clearly changed by 1929, when Hubble published his first paper on the velocity-distance relation [30]. Only a few radial velocities had been added since the mid-1920s, with 46 measurements available to Hubble. These were combined with individual distances to 24 galaxies, mostly using data from the 1926 paper, and based on observations of (what Hubble assumed to be) individual supergiant stars. The scatter in the resulting velocity-distance relation remained considerable, but Hubble's somewhat cautious remark that “For such scanty material, so poorly distributed, the results are fairly definite.” nevertheless seems justified. The highest radial velocity remained the 1800 km/s of NGC 584, and it was clear that the best way to establish a more convincing velocity-distance relation was to push the measurements to greater distances, and hence higher velocities if the relation was real. Such efforts were already underway at Mount Wilson.

Overall, the 1929 paper has the character of a relatively brief preliminary report (the paper is only 6 pages long, including the tables with velocity and distance measurements and the figure showing the velocity-distance relation). Perhaps Hubble felt some urgency to publish the results, and one can discuss whether he gave due credit to earlier work. He did briefly mention Lundmark's 1925 paper (with the quadratic fit to the velocity-distance relation) but not the 1924 paper in which the velocity-distance relation was plotted. Van den Bergh [31] has suggested that Hubble may have downplayed Lundmark's work because of a strained personal relationship between the two astronomers (Hubble had accused Lundmark of plagiarizing his galaxy classification scheme [25]). However, Wirtz's papers were not mentioned either. Although these were published in German, this would hardly have been a hindrance for Hubble, who was quite proficient in that language [31].

One may also wonder about the lack of reference to the previous determinations of the slope of the relation by Lemaître and Robertson. Neither of these papers was published in the mainstream literature, and it is possible that Hubble had not read them. However, Humason later recalled that Hubble had come back from the IAU General Assembly in 1928 in Leiden, “rather excited about the fact that two or three scientists over there, astronomers, had suggested that the fainter the nebulae were, the more distant they were and the larger the redshifts would be.”. According to van den Bergh, some of the astronomers that Hubble would likely have met included de Sitter, Lemaître, Lundmark, Shapley, and Smart [31]. Furthermore, Robertson had obtained his PhD in 1925 from Caltech and worked there as an assistant professor from 1927-1928 before moving to Princeton. So it appears likely that Hubble was aware of the work by both Lemaître and Robinson. On the other hand, their contributions were primarily theoretical, and theoretical considerations were delegated to only a brief paragraph at the end of Hubble's 1929 paper. He remarked: “The outstanding feature, however, is the possibility that the velocity-distance relation may represent the de Sitter effect, [...]. In the de Sitter cosmology, displacements of the spectra arise from two sources, an apparent slowing down of atomic vibrations and a general tendency of material particles to scatter.”. This interpretation refers to the original version of de Sitter's model, and Hubble's formulation is reminiscent of the wording in Eddington's book: “De Sitter's theory gives a double explanation of this motion of recession: first, there is the general tendency to scatter ... second, there is the general displacement of spectral lines to the red in distant objects due to the slowing down of atomic vibrations.”. One gets the impression that Hubble had some familiarity with de Sitter's model - by far the most established at the time - and felt compelled to mention it, but that he did not consider it important to dwell on the differences between that and the more recent models by Lemaître and Robertson.

The omission of reference to Lemaître is somewhat more conspicuous in Hubble's book ``The realm of the nebulae'', published in 1936 [32]. Here Hubble reviewed the early observational work on the velocity-distance relation by Wirtz (“the leader in the field”), Lundmark, Strömberg, and others in some detail. He also mentioned the theoretical work by de Sitter, Friedman, and Robertson, with a reference to “Robertson's authoritative review” from 1933. Robertson's review discussed Lemaître’s work to the extent that it could hardly be overlooked, although it did not mention the comparisons with observations done by either Lemaître or Robertson himself. By 1936, of course, the English translation of Lemaître‘s 1927 paper had appeared, but without the paragraph on the radial velocity measurements.

From 1930 onwards, the existence of the linear velocity-distance relation rapidly became an established fact. New radial velocity measurements for more distant galaxies were being obtained by Humason at Mount Wilson with the help of a new fast spectrograph camera employing a Rayton lens at the Cassegrain focus of the 100-inch telescope. Using these data, de Sitter extended the velocity-distance relation to about 8000 km/s in 1930, confirming a clear correlation [33].  De Sitter mentioned Lemaître‘s 1927 paper and wrote that “I hope to return to the discussion of this ingenious solution in a separate communication.” By 1931, Humason had added 46 new radial velocity measurements, with the highest velocities extending to 19700 km/s [34]. Together with distance estimates based on the integrated magnitudes of the brightest cluster galaxies, these measurements firmly established the linear velocity-distance relation beyond any doubt, although the inferred slope remained much too steep [35].

It is well known that Hubble's distances were underestimated for a variety of reasons. Most of the objects that he assumed to be supergiant stars were not actually individual stars, but HII regions, as demonstrated in 1956 by Humason, Mayall, & Sandage [36]. The possibility that the brightest sources were not individual stars was acknowledged in the Hubble & Humason 1931 paper (“The use of the brightest stars as a criterion of distance has been criticized on the ground that clusters and groups at such remote distances would not be distinguished from single stars.”), and again in The realm of the nebulae, although it was dismissed perhaps a bit too easily. Another important issue was the calibration of Cepheid period-luminosity relation. This is a story on its own, as discussed by Baade [37]. Nevertheless, taking into account both of these issues, the value of the Hubble constant determined by Humason et al.\ in 1956 was still 180 km/s/Mpc, more than twice the current best estimate. Unlike the radial velocities, which follow directly from the spectra and were basically correct from the beginning, it took a long time to get the distances right, and establishing the value of the Hubble constant to ever greater precision remains a work in progress.

On the theoretical side, it became clear from the wide range of possible models that measuring the local slope of the velocity-distance relation alone was insufficient for determining the curvature of the Universe. Einstein & de Sitter succinctly summarised the situation in 1932: “…we must conclude that at the present time it is possible to represent the facts without assuming a curvature of three-dimensional space. The curvature is, however, essentially determinable, and an increase in the precision of the data derived from observations will enable us in the future to fix its sign and to determine its value.” [38].

 

Concluding remarks

 

It should be evident that there is no unique answer to the question of who discovered the expansion of the Universe. The velocity-distance relation emerged gradually during the 1920s through the efforts of several authors, with theoretical and observational work progressing in parallel. Evidence of the predominantly positive radial velocities of spiral nebulae was available even before the formulation of GR and was known to theorists. Establishing the linear relationship with distance took longer, but this did not deter several authors from using the available data to estimate the curvature of the Universe. In the context of de Sitter's model, this was equivalent to determining the slope of the distance-velocity relation (what we now call the Hubble constant), although it was not usually formulated as such. While de Sitter's model implied an expanding Universe, this was not immediately recognised.

Hubble's 1929 paper was one of several important milestones, but more important than the paper itself was the resulting extensive follow-up programme that was initiated at Lick, Mount Wilson, and Palomar to extend the radial velocity measurements to greater distances, thereby corroborating the earlier results.  By 1931, no doubt could remain about the existence of a linear distance-redshift relation, and the programme culminated with the publication of 800 radial velocities by Humason et al. in 1956, extending to 60 000 km/s. In the introduction, the authors paid tribute to Hubble (who had passed away three years earlier), writing: “These spectrographic and photometric data are now available in considerable numbers ... chiefly as the result of Hubble's inspiring influence on his colleagues.”

The importance of the work by Lemaître and Robertson (as well as Friedman) lay primarily in their theoretical contributions, which marked a departure from de Sitter's model - the only model which, up until then, could account for the radial velocity measurements. Both authors gave the modern interpretation of the velocity-distance relation as a signature of an expanding Universe, and provided quantitative estimates of the slope of the relation prior to Hubble's 1929 paper. However, it is difficult to argue that they demonstrated the existence of the linear distance-redshift relation much more convincingly than Wirtz or Lundmark before them (neither Lemaître nor Robertson actually published plots of the distance-velocity relation).

With the realisation that the possible solutions to the field equations were not restricted to those of Einstein and de Sitter, it became clear that determining the curvature of space required measuring the departures of the velocity-distance relation from linearity. Many of the main observational tests to accomplish this were laid out in detail by Sandage in 1961 [39], but definitive results were only obtained in the last decade of the 20th century. It is somewhat ironic that, after a century of attempts to determine the curvature, it remains “possible to represent the facts without assuming a curvature of three-dimensional space.”

 

 

References

 

[1]      V. M. Slipher, “The radial velocity of the Andromeda Nebula,” Lowell Obs. Bull., vol. 1, 1913.

[2]      J. Scheiner, “On the spectrum of the great nebula in Andromeda.,” Astrophys. J., vol. 9, p. 149, Mar. 1899.

[3]      V. M. Slipher, “Spectrographic Observations of Nebulae,” Pop. Astron., vol. 23, pp. 21–24, 1915.

[4]      V. M. Slipher, “Radial velocity observations of spiral nebulae,” Obs., vol. 40, pp. 304–306, 1917.

[5]      A. Einstein, “Kosmologische Betrachtungen zur allgemeinen Relativitätstheorie,” Sitzungsberichte der Königlich Preußischen Akad. der Wissenschaften, p. 142, 1917.

[6]      W. de Sitter, “On Einstein’s Theory of Gravitation and its Astronomical Consequences. Third Paper.,” Mon. Not. R. Astron. Soc., vol. 78, no. 1, pp. 3–28, Nov. 1917.

[7]      K. Lundmark, “Die Stellung der kugelförmigen Sternhaufen und Spiralnebel zu unserem Sternsystem,” Astron. Nachrichten, vol. 209, no. 24, pp. 369–380, 1919.

[8]      H. Weyl, “Zur allgemeinen Relativitätstheorie,” Phys. Zeits., vol. 24, pp. 230–232, 1923.

[9]      A. S. Eddington, The mathematical theory of relativity. Cambridge University Press, 1923.

[10]    L. Silberstein, “The Curvature of de Sitter’s Space-Time derived from Globular Clusters,” Mon. Not. R. Astron. Soc., vol. 84, no. 5, pp. 363–366, Mar. 1924.

[11]    C. Wirtz, “Über die Bewegungen der Nebelflecke,” Astron. Nachrichten, vol. 206, no. 13, pp. 109–116, 1918.

[12]    C. Wirtz, “Notiz zur Radialbewegung der Spiralnebel,” Astron. Nachrichten, vol. 216, no. 24, pp. 451–452, 1922.

[13]    C. Wirtz, “De Sitters Kosmologie und die Radialbewegungen der Spiralnebel,” Astron. Nachrichten, vol. 222, no. 2, pp. 21–26, 1924.

[14]    C. Wirtz, “Ein literarischer Hinweis zur Radialbewegung der Spiralnebel,” Zeitschrift für Astrophys., vol. 11, p. 261, 1936.

[15]    W. C. Seitter and H. W. Duerbeck, “Carl Wilhelm Wirtz - Pioneer in Cosmic Dimensions,” ASP Conf. Ser., vol. 167, p. 341, 1999.

[16]    K. Lundmark, “The determination of the curvature of space-time in de Sitter’s world,” Mon. Not. R. Astron. Soc., vol. 84, p. 747, 1924.

[17]    K. Lundmark, “The Motions and the Distances of Spiral Nebulae,” Mon. Not. R. Astron. Soc., vol. 85, no. 8, pp. 865–894, Jun. 1925.

[18]    G. Stromberg, “Analysis of radial velocities of globular clusters and non-galactic nebulae.,” Astrophys. J., vol. 61, p. 353, Jun. 1925.

[19]    A. Friedman, “Über die Krümmung des Raumes,” Zeitschrift für Phys., vol. 10, no. 1, pp. 377–386, Dec. 1922.

[20]    M. Heller, “Friedman’s Cosmological Views,” Acta Cosmol., vol. 13, p. 65, 1985.

[21]    H. Nussbaumer, “Einstein’s conversion from his static to an expanding universe,” Eur. Phys. J. - Hist., vol. 39, pp. 37–62, 2014.

[22]    G. Lemaitre, “Note on de Sitter’s Universe,” J. Math. Phys., vol. 4, pp. 188–192, 1925.

[23]    S. Mitton, “Georges Lemaitre: Life, Science and Legacy.” 2016.

[24]    G. Lemaître, “Un Univers homogène de masse constante et de rayon croissant rendant compte de la vitesse radiale des nébuleuses extra-galactiques,” Ann. la Société Sci. Bruxelles, pp. 49–59, 1927.

[25]    E. P. Hubble, “Extragalactic nebulae.,” Astrophys. J., vol. 64, p. 321, Dec. 1926.

[26]    G. Lemaitre, “A Homogeneous Universe of Constant Mass and Increasing Radius accounting for the Radial Velocity of Extra-galactic Nebulae,” Mon. Not. R. Astron. Soc., vol. 91, no. 5, pp. 483–490, Mar. 1931.

[27]    M. Livio, “Mystery of the missing text solved,” Nature, vol. 479, no. 7372, pp. 171–173, Nov. 2011.

[28]    H. P. Robertson, “On relativistic cosmology,” London, Edinburgh, Dublin Philos. Mag. J. Sci., vol. 5, pp. 835–848, 1928.

[29]    H. P. Robertson, “On the foundations of relativistic cosmology,” Proc. Natl. Acad. Sci., vol. 15, pp. 822–829, 1929.

[30]    E. Hubble, “A relation between distance and radial velocity among extra-galactic nebulae,” Proc. Natl. Acad. Sci., vol. 15, no. 3, pp. 168–173, Mar. 1929.

[31]    S. van den Bergh, “Discovery of the Expansion of the Universe,” J. R. Astron. Soc. Canada, vol. 105, p. 197, Aug. 2011.

[32]    E. P. Hubble, The realm of the nebulae. New Haven and London: Yale University Press, 1936.

[33]    W. de Sitter, “On the magnitudes, diameters and distances of the extragalactic nebulae and their apparent radial velocities,” Bull. Astron. Institutes Netherlands, vol. 5, p. 157, 1930.

[34]    M. L. Humason, “Apparent Velocity-Shifts in the Spectra of Faint Nebulae,” Astrophys. J., vol. 74, p. 35, Jul. 1931.

[35]    E. Hubble and M. L. Humason, “The Velocity-Distance Relation among Extra-Galactic Nebulae,” Astrophys. J., vol. 74, p. 43, Jul. 1931.

[36]    M. L. Humason, N. U. Mayall, and A. R. Sandage, “Redshifts and magnitudes of extragalactic nebulae.,” Astron. J., vol. 61, p. 97, Apr. 1956.

[37]    W. Baade, “The Period-Luminosity Relation of the Cepheids,” Publ. Astron. Soc. Pacific, vol. 68, p. 5, Feb. 1956.

[38]    A. Einstein and W. de Sitter, “On the Relation between the Expansion and the Mean Density of the Universe,” Proc. Natl. Acad. Sci., vol. 18, no. 3, pp. 213–214, Mar. 1932.

[39]    A. Sandage, “The Ability of the 200-INCH Telescope to Discriminate Between Selected World Models.,” Astrophys. J., vol. 133, p. 355, Mar. 1961.