Fig. 4.21. Measures of the background radiation spectrum and calculations of the helium abundance in December 1964.
of starlight, an important advance that required a lot of work (Hauser and Dwek 2001). The point at the left edge of the graph shows a measured upper bound on the background radiation at about 1-m wavelength. The energy in the radio background contributed by the measured counts of radio sources was known; I added that to the version of the figure I showed in my second colloquium.
I think the upper limit at microwave wavelengths in Figure 4.21 refers to a Bell Laboratories paper we were discussing in the Gravity Group, Hogg and Semplak (1961). We interpreted it as giving an upper bound of about 15 K. If so, I made a mistake in the figure: the wavelength is 5 cm, not 3 cm. I don't know why we overlooked the better reference, Ohm (1961). And we had not yet noticed - and pointed out to Bob Dicke - that he had placed a bound T< 20 K at 1-cm wavelength (Dicke et al. 1946). It is at these microwave wavelengths that fossil thermal radiation might appear: not so hot as to have an unacceptably large energy density for the relativistic big bang cosmology nor so cool as to be unobservable.
I later learned that Doroshkevich and Novikov (1964) had made a similar study of the cosmic radiation energy density. Their version of the CMBR spectrum is shown in Figure 4.5. The focus of their analysis was the mean energy density as a function of wavelength from the accumulated amounts of starlight and radio radiation produced by the galaxies. But they remark that the "Gamow theory" would produce a thermal spectrum at microwave wavelengths, and they refer to the paper by Ohm (1961) on the Bell Laboratories communications experiments discussed in Chapter 3 and by Hogg, Penzias, and Wilson in this chapter. It is a better bound than the one I showed.
The right-hand graph in Figure 4.21 shows my computation of the mass fraction X left in hydrogen at the end of the big bang thermonuclear reactions discussed in Chapter 3. Almost all of the rest of the baryons, with mass fraction Y = 1 — X, would be in helium. I did not compute the production of heavier elements, but felt it would be small. The arrow on the left is an estimate of X in the interstellar plasma, and the line on the right is my guess of a reasonable lower bound on X in the earliest generations of stars. The horizontal axes show the present radiation temperature computed for two possible values of the present mean mass density (in baryons; I wasn't thinking about nonbaryonic dark matter). According to my notes for the Wesleyan talk I pointed out that an interesting value of Y in a universe with a hot big bang could be associated with microwave background radiation that would be warm enough to be detectable. (Doroshkevich and Novikov 1964 also very clearly made that point.) I don't remember whether I mentioned the Princeton experiment aimed at its possible detection.
We were of course thinking about how we might interpret the experiment if it were found that there is undetectably little microwave radiation. According to the assumptions used in the figure the low temperature would imply an unacceptably low value of X (a large helium abundance). But there are ways out; I mentioned two at Wesleyan (according to my notes, which were my security blanket in those days). The first is that the big bang may have been cold. That can be reconciled with large X by postulating that there are enough neutrinos to prevent formation of neutrons. This is what is meant by the comment about leptons in the figure. I noticed only later that Zel'dovich (1962, 1965) had independently argued for a cold big bang. His reasoning, discussed on page 35, led him to the proposal that the early universe contained equal numbers of baryons, electrons, and neutrinos, that is, the lepton number is equal to twice the baryon number. That is all the leptons needed to eliminate helium production in a cold big bang.
The second way out I mentioned is that the universe is not even approximately homogeneous and isotropic: maybe there was no big bang. The homogeneity assumption is well supported by the observations now, but the evidence was sparse then.
In the paper Dicke and Peebles (1965) (which was submitted before we knew about Penzias and Wilson, but it has a comment added in proof) we mention a third possibility, that general relativity theory is not valid.
It is after all an enormous extrapolation of the theory from the meager tests we had then to its application on the scales of length and time of the expanding universe. At the time, Bob Dicke was very interested in the idea discussed on page 37. Perhaps the strength of the gravitational interaction was large in the early stages of expansion of the universe, maybe comparable to the strength of the electromagnetic interaction. Maybe the gravitational interaction is weak now because it has been decreasing for a long time. In this picture the rate of expansion and cooling of the early universe could have been rapid enough to have prevented significant element building.
We should have mentioned yet another possibility: in the steady state cosmology the continual creation of matter could have included helium or microwave radiation in large or small amounts: it's a free assumption. As I said earlier, my recollection is that at that time neither of us found the steady state philosophy interesting.
My notes for the Wesleyan talk suggest I had nothing useful to say about the astronomical determinations of the cosmic helium abundance Y. But at Princeton we were learning that Y is larger than seems likely to be accounted for by production in stars, and maybe is in line with a hot big bang. We mention that in Dicke and Peebles (1965), with a reference we had just learned to Hoyle and Tayler (1964), who knew a lot more about the astronomy than we did.
Bob and I did not refer to Osterbrock and Rogerson (1961). I now believe they were the first to present evidence for a critical result: the older stars that contain fewer heavy elements appear to contain about the same helium mass fraction as younger stars. They also point out the possibly very significant implication, that "the build-up of elements to helium can be understood without difficulty on the explosive formation picture (Gamow 1949)." I only remembered when writing this essay that I had referred to Osterbrock and Rogerson in a study of the structure of the planet Jupiter (Peebles 1964). I concluded that a small fraction of Jupiter's mass is in heavy elements, most of which had settled to a central core, while most of the mass outside the core is a mixture of hydrogen and helium. The mass fraction in helium that fit Jupiter's mass, radius, and rotational flattening is roughly consistent with the Osterbrock and Rogerson estimates, and with what Martin Schwarzschild told me about the composition of the Sun. The remark by Osterbrock and Rogerson about the explosive formation picture would not have meant much to me when I was making models of Jupiter: I knew close to nothing about cosmology. When I started thinking about a hot big bang I should have remembered the evidence for large Y in old stars. Bob Dicke liked to say that "we get too soon old and too late smart."
My second colloquium on cosmology was at the Applied Physics Laboratory at the Johns Hopkins University in Maryland, on February 19, 1965. I don't know why I was invited; maybe it had something to do with the fact that Alpher and Herman were at the Applied Physics Laboratory when they were developing the physics of element production in a hot big bang. But I learned about that connection much later.
In this colloquium I presented updated versions of the helium production calculation and the cosmic radiation spectrum. I had added to the latter a bound on the energy in microwave radiation from the absence of a discernible effect of its drag on energetic cosmic ray protons. That pretty convincingly ruled out the idea that the mass of the universe might be dominated by thermal radiation. But it did leave room for an interesting fossil thermal background.
I had asked David Wilkinson whether it would be appropriate to mention the Roll-Wilkinson experiment. You want people to know about your work, but only when it is unlikely someone else might be inspired to do it first. Dave assured me that no one could catch up with them at that point, so I mentioned the experiment. Ken Turner, a friend since our graduate student days in the Gravity Group, attended the talk. He told Bernie Burke about it. Burke brought the news to Arno Penzias and Bob Wilson. They were at the Bell Laboratories in Holmdel, NJ, not far from Princeton. They did not have to catch up: they had already done the experiment. Arno telephoned the news to Bob Dicke.
What was our reaction to the telephone call? I remember relief and excitement: they showed us that there actually is something to be measured, always a very good thing. That overwhelmed any chagrin over priority, and to me it still does, with one exception. The Nobel Prize rightly went to Penzias and Wilson: they made very sure of the reality of an unexpected result, and they made sure the world knew about it. But the Nobel committee should have included Dicke.
When and how did I learn that my first computations of light element formation largely repeated earlier work? My records reveal a few data points. I submitted a paper on my calculations to the journal Physical Review. The referee recommended rejection, saying that my calculations had already been done, and by whom. I revised and resubmitted several times. I have a draft dated January 1965 that has a reference to Alpher and Herman (1953), but I don't know whether this draft was the first to recognize that I was repeating old analyses. I have a copy of a letter I wrote to Hoyle and Tayler on February 1, 1965, acknowledging their prior work. I have a copy of my letter to Physical Review dated June 23, 1965, in which
I at last withdrew the paper. By then I had faced up to the fact that to make a meaningful contribution I would have to do a distinctly better computation.
Fred Hoyle also saw the need for a better computation, and he, Willy Fowler, and Bob Wagoner got to work. I met Bob Wagoner at a conference in Miami in December 1965. We exchanged ideas but not techniques of computation. I devised fixes for the numerically unstable reaction equations that work but likely would be close to incomprehensible to anyone else. Wagoner used the more familiar (to Fowler and Hoyle) techniques from the analyses of nuclear burning in stars, and he considered a larger set of nuclear reactions. I believe their computer code evolved into some that are used today. But the important elements of our results, in Peebles (1966) and Wagoner, Fowler and Hoyle (1967), agree.
Our ignorance in 1964 about the literature of this subject is legendary in the cosmology community, and legends beguile. I see the effect in Bob Dicke's comment (unpublished, dated 1975):
There is one unfortunate and embarrassing aspect of our work on the fire-ball radiation. We failed to make an adequate literature search and missed the more important papers of Gamow, Alpher and Herman. I must take the major blame for this, for the others in our group were too young to know these old papers. In ancient times I had heard Gamow talk at Princeton but I had remembered his model universe as cold and initially filled only with neutrons.
I think Bob apologized too much. I have the greater share of blame for poor homework: Bob was careful to stand back and let younger people in his group get on with research on their own. Our paper Dicke and Peebles (1965) did not give proper references to earlier work on the hot big bang, but we remedied that pretty quickly. I believe the citations are normal and proper in Dicke et al. (1965), the paper that offers the hot big bang interpretation of the Penzias and Wilson (1965a) detection. Because I have on occasion encountered the myth that our paper did not refer to earlier work I list our relevant references: Alpher, Bethe and Gamow (1948), Alpher, Follin and Herman (1953), and Hoyle and Tayler (1964). I don't remember whether the absence of a reference to the prediction of the CMBR in the present universe by Alpher and Herman (1948) signifies more than lack of careful reading. And the list of references is brief, but then this is a brief paper. In the late 1960s Dave Wilkinson and I systematically advertised the history of ideas in our lectures at conferences and colloquia; one sees an example in Figure 3.1 at the start of Chapter 3. And I think there is a full and accurate account of the history in Chapters V and VIII of Physical Cosmology (Peebles 1971).
Bob hated sloppy physics, a term he used on occasion to express strong disapproval. I don't remember his ever applying those feared words to me, though I do remember clear reprimands for less than careful physics. My homework in 1964 could be termed "sloppy," but I don't remember Bob or anyone else in the group chiding me about it then or later. We were caught up in the excitement of exploring rich and sparsely worked ground.
The other rich slice of physics I started pursuing in 1964 is the effect of the CMBR on the gravitational growth of small initial departures from an exactly homogeneous mass distribution into the present strong clustering of mass on the scale of galaxies. Here again Gamow (1948a) got there first. He pointed out that the matter temperature and density in the early universe determine the pressure, and the pressure sets the size of the smallest cloud of matter that gravity can cause to break away from the general expansion. This is the analog in cosmology of the Jeans criterion for the balance of gravitational attraction and the pressure gradient force of repulsion of a cloud of matter. Gamow also argued that the gravitational instability to the growth of mass clustering commences when the mass density in matter becomes larger than that in radiation. He was right, though his argument is not what we use today. A brilliant physicist can do that.
I was able to add something new. I found that when the universe was young and hot enough to ionize the baryons the drag of the radiation on the plasma is strong enough to prevent the gravitational formation of a non-relativistic cloud of baryons. That situation changes when the temperature dropped to about 3000 K, the plasma combined to largely atomic hydrogen and helium, and matter and radiation abruptly decoupled. I published the idea in Peebles (1965) (with, I am relieved to see, appropriate reference to Gamow 1948a on the Jeans length).
The departures from an exactly homogeneous mass distribution must disturb the spectrum and spatial distribution of the radiation. Here are my recollections of how this was worked out in the 1960s.
At the January 1967 Texas Symposium on Relativistic Astrophysics in New York City I presented a more detailed analysis of the behavior of the matter-radiation fluid prior to decoupling in general relativity theory. I took account of the effective viscosity from the diffusion of radiation through the plasma, and analyzed how the dissipation suppresses small-scale density fluctuations that act as pressure waves. I presented my paper on these considerations for publication in the conference proceedings, but because of turmoil at the publisher the proceedings never appeared in print. Richard Michie (1967) independently worked out main elements of this physics, but illness prevented this work from getting past the preprint stage. Joe Silk also independently worked it out, and he published (Silk 1967, 1968b), so the effect is properly termed "Silk damping."
Sachs and Wolfe (1967) derived the gravitational perturbation that, in general relativity theory, dominates the large-scale disturbance to the space distribution of fossil radiation from the big bang. Chapter 5 traces how that deeply influential relation was detected a quarter of a century later.
The acoustic (sound wave) oscillation of the matter-radiation fluid can leave characteristic patterns - that much later actually were measured -in the distributions of matter and radiation. In his contribution Sunyaev describes how he and Zel'dovich (1970c) derived analytic approximations to the analysis of this effect. I didn't know what they were doing until quite a while later; communications with people in the USSR were slow. But as it happened I took the next step (while enjoying sabbatical leave and the hospitality of Caltech in 1968-1969) of working out the radiative transfer analysis that is needed for a more complete computation of the residual patterns in the distributions of matter and radiation. Jer Yu, who had been my first graduate student, joined me in the numerical solutions (Peebles and Yu 1970). Our paper shows the power spectrum representation of the relict acoustic oscillations in the mass distribution, along with a pretty awkward way to represent the acoustic oscillations in the radiation distribution. Apart from that - and the dark sector, and the unsophisticated numerical methods - this is close to the standard physics used in the analyses of the measurements of the variation of the CMBR temperature across the sky, including the wonderfully precise WMAP data discussed in the next chapter.8
The CMBR also is disturbed by its interaction with plasma that in the present-day universe is hotter than the radiation: scattering by the hot free electrons pushes the spectrum of the radiation down from blackbody at long wavelengths and up at the short-wavelength end. The plasma in clusters of galaxies is particularly hot and dense. It produces an observable disturbance to the CMBR that has become a useful diagnostic of how cosmic structure formed. Ray Weymann was among the first to analyze this important effect in a hot big bang cosmology. He writes:
I then became interested in understanding the coupling (and subsequent decoupling) of the matter and radiation and came to the realization that the Compton
8 Another piece of the physics is the combination of the primeval plasma into mostly neutral atomic hydrogen and helium, with small but important amounts of free electrons and molecular hydrogen. The computations are presented in Peebles (1968) and Zel'dovich, Kurt and Sunyaev (1968). Rashid Sunyaev explains the computation and how I learned about their independent work. There are quite a few other examples of this parallel development of ideas in the 1960s. I think it's not surprising: once people became conscious of the radiation these became pretty obvious things to work out.
interaction was the dominant interaction mechanism. To derive the frequency-dependent interaction and its diffusion approximation, one does need special relativity, and I was helped by a paper and correspondence with Willard Chappell in Boulder, Colorado, who helped me over an obstacle. The resulting paper was published in Physics of Fluids (Weymann 1965). Not very long after that I applied that diffusion equation to study the temperature history of the matter and radiation, and that involved studying the recombination era. I wrote up two papers, and I believe one was published (after a struggle with the referee; Weymann 1966) but the other only appeared as a Steward Observatory preprint. One of these papers calculated departures from the Planck function that would result under various (and I later realized mostly unrealistic) heating mechanisms.
Shortly after this I received a letter from Zel'dovich who pointed out that the diffusion equation I had derived had already been derived by Kompaneets, though I was totally unaware of it, as it was in a Soviet journal. About then the Zel'dovich and Sunyaev (1969) paper came out. If you read that paper you will see that my paper was referenced fairly extensively by them, but my paper had two serious defects: I did not derive the analytic expression which they did, but relied only on numerical work, and I only applied the work to the cosmic expansion and not to finite clouds of electrons.
My only regret in all this is that since I did all that work at Arizona, there was at that time no other theoretician there to talk to and so I was too isolated and the work I did there suffered from that.
I had been thinking about this effect too, and had worked out the theory by 1970, but that was after Sunyaev and Zel'dovich. I at first didn't much care that I was scooped because I didn't think the effect would be large enough to be observable. Dave Wilkinson straightened me out on that, though a convincing detection did take a lot of work.
These analyses assumed the microwave radiation really is a fossil from the very early universe. An alternative that had to be considered in the 1960s was that the CMBR was produced by sources in the universe as it is now or at modest redshifts. Galaxies are sources of optical and radio radiation; might they also produce the microwave background? This local source model was discussed by Sciama (1966), Gold and Pacini (1968), Wolfe and Burbidge (1969), and Pariiskii (1968). It was a serious possibility in the 1960s that demanded tests: measurements of the spectrum and angular distribution of the radiation. If a fossil from the hot big bang the radiation spectrum ought to be close to blackbody. That spectrum would not likely be produced by microwave sources at low redshift, and at adequate angular resolution the radiation would break up into the individual sources. Layzer (1968), and later Hoyle, Burbidge and Narlikar (1993), postulated that absorption and reradiation by dust relaxed the CMBR spectrum toward blackbody, and smoothed out the radiation. The picture didn't seem promising, though, because we knew distant radio-emitting galaxies are observed at CMBR wavelengths, with no indication of absorption.
My notes for a colloquium on March 17, 1966, at the University of Toronto show significant advances toward the measurement of the spectrum. In addition to the Penzias and Wilson detection at wavelength A = 7.4 cm I could show the very recently published Roll and Wilkinson (1966) measurement at 3.2 cm and the CN temperature measurement (as in equation 3.13) at 2.6mm by Field, Herbig and Hitchcock (1966). The fit to a thermal spectrum certainly looked promising. And there was another data point, the consistency with the helium abundance. By this time I was arguing that we had a significant case for the hot big bang model, and a serious challenge therefore for the steady state cosmology.
I can recall early reactions by a few others to our proposed interpretation of the CMBR. In July 1965 Bob Dicke and I attended a conference on general relativity and gravitation at Imperial College, London. The microwave radiation was not on the program but there were informal discussions. We met Fred Hoyle; I remember the talk as friendly, but, for whatever reason, short. I don't remember meeting Igor Novikov, but he recalls (p. 99) bringing the news of the microwave radiation back to Zel'dovich.
Bob Dicke showed us a letter he received from Zel'dovich. In this letter, dated September 15, 1965, Zel'dovich writes
I am not more so cock-sure in my colduniverse hypothesis: It was based on the assumption that the initial helium content is much smaller than 35% by weight. Now I understand better the difficulty of helium determination. You draw some conclusions from the observed helium content 25%. Are you sure it is not 35% or 15%?
It seems to me very desirable to measure the Planck spectrum corresponding to 3—4°K at its maximum, at the wave-length ~1mm, although it is a difficult task.
Undoubtedly your work will raise the interest to all sides of the problem and I sincerely congratulate you and your team on a success.
Zel'dovich went on to argue that Dicke's oscillating universe picture is "untenable as a consequence of unlimited growth of entropy." We knew the argument, but I think I recall that we were not so sure that entropy need be conserved in the bounce. This was before the full development of the cos-mological singularity theorems Ellis discusses (commencing on p. 379), but we were aware of the general idea. I remember Bob saying, in effect, that general relativity predicts that a collapsing universe develops a singularity but it doesn't say whether the singularity applies to the whole universe or just a bit of it, maybe leaving a few black holes, while the rest of the universe expands again. What would happen to the accumulation of black holes as well as entropy over many cycles? I was inclined to work on something else.
In his letter of reply to Zel'dovich, dated October 5, 1965, Bob suggests that "the helium content of the proto-galaxy could very well have been zero."
That was inspired by his fascination with the possibility that the strength of the gravitational interaction decreases as the universe expands. If so it would make the rate of expansion of the early universe much larger than in the standard model. If the expansion were fast enough there would be negligible light element production at high redshift. I don't remember whether I told Bob about the Osterbrock and Rogerson (1961) evidence for large helium in relatively old stars.
At the time of this letter there was in the literature a direct measurement of the microwave radiation at just one wavelength, and one indirect measurement from the interstellar molecule CN. Zel'dovich seemed to be ready to accept that the spectrum likely will prove to be blackbody, but quite a few others reminded me that that is a considerable extrapolation from the measurement of one (or two if you trust CN) point on the spectrum. I remember a conversation with Phil Morrison in a noisy room. He said, in effect, measure the energy of the sound in this room and convert it to an effective temperature. You'll get an absurd value. He bet one guinea that the same is true of the microwave radiation, that measurements at other wavelengths would not follow the thermal spectrum. I think it was at the 1967 Texas Symposium that he agreed that he had likely guessed wrong and paid me one pound and one shilling. I met Ralph Alpher at that meeting, for the only time, but our conversation was short, I think because I was rushing to catch the train home.
By the end of 1966 Howell and Shakeshaft (1966) and Penzias and Wilson (1967) had added a data point at 21-cm wavelength to the measurements at 7 cm, 3 cm and 2.6 mm. The spectrum up to the expected peak looked encouragingly close to blackbody. Not long after this measurement, in a letter dated December 21, 1966, Dennis Sciama wrote to Bob Dicke,
As you may have heard I have recanted from the steady state theory, and have taken such a liberal dose of sackcloth and ashes that I am now more orthodox than the orthodox (though I don't suppose this phase will last long). Anyway you can tell Peebles that I now nearly believe that the excess background has a black body spectrum. I hope to see him and you in New York so that I can capitulate in person.
Sciama's new phase did last: he continued to work on the relativistic big bang cosmology, with particular attention to clues to the physics of the dark matter (Sciama 2001).
By 1970 three groups had attempted to measure the CMBR energy spectrum at wavelengths near 1 mm, where the spectrum is expected to break away from the power-law form that applies at longer wavelengths. As Zel'dovich had remarked, and Harwit (p. 329) and Weiss (p. 342) describe, that "is a difficult task." From 1970 to 1990 a series of experiments indicated that the CMBR spectrum significantly differs from blackbody near and shortward of the blackbody peak. The beautiful experiments by Mather et al. (1990) and Gush, Halpern and Wishnow (1990) at last showed that the spectrum is wonderfully close to thermal.
I have no complaints about the two-decade-long apparent anomaly in the spectrum - we were seeing first-rate science in progress - but it did confuse the subject and it led me to think about other things. That mainly was the statistical analyses of the clustering and the dynamical analyses of the motion of matter on large scales. At the time that was a better subject for me to work on anyway. It is the sort of thing I like doing, the field was ripe for exploration, and it grew into a component of the second critical test of the cosmological interpretation of the CMBR, the signature in its variation across the sky of its interaction with the growing inhomogeneity in the mass distribution.
I can date my work on measures of the cosmic clustering of matter to the March 1966 colloquium in Toronto. Sidney van den Bergh asked me how I could be sure the universe really is close to homogeneous in the large-scale average. I offered as evidence the CMBR, which we already knew is quite smooth, consistent with a near uniform large-scale mass distribution. The argument is pretty indirect, of course. Sidney countered that George Abell's map of the distribution of rich clusters of galaxies (Abell 1958) does not look very smooth. I said it doesn't look all that rough, considering the sparse sampling. I think I can remember Sidney's words, "you could check that." I worked out a method of checking it on the flight back home, and Jer Yu improved and applied it in his PhD thesis (Yu 1968; Yu and Peebles 1969).
The obvious measures for this project are second moments: the two-point correlation function and its transform, the power spectrum. My initial choice of the latter was influenced by Bob's preference and by the arguments in The Measurement of Power Spectra (Blackman and Tukey 1958). That proved to be right for measurements of the large-scale distributions of matter and radiation, but I learned that correlation functions are better suited to measures of the nonlinear clustering of matter on relatively small scales.
I continued the analysis of statistical measures of the distributions of extragalactic objects - n-point correlation functions and their transforms -and of the dynamical evolution that might produce the observed clustering, for more than a decade. There was a positive reason: this was rich fallow ground to explore. And there was a negative one: I mentioned the spectrum anomaly that beclouded my thoughts about the CMBR. Though I like to work alone, I needed help in this data analysis and interpretation, and it appeared. Along with Jer Yu, I am deeply grateful (though it may not have always been apparent at the time) for collaborations on these statistical analyses with Martin Clutton-Brock, Marc Davis, Jim Fry, Margaret Geller, Ed Groth, Mike Hauser, Dan Hawley, Bernard Jones, Diego Lambas, Mike Seldner, Bernie Siebers, and Raymond Soneira. All were volunteers. Young people somehow tend to sense when and where things of possible interest are happening.
In 1969 I gave a graduate course at Princeton on current topics of research in cosmology. John Wheeler insisted that I turn the course into a book, and he sat in the back of the room and took notes until I agreed. That so unnerved me that I wrote Physical Cosmology (Peebles 1971). By then I understood that cosmology is a real physical science that offers fascinating issues of theory and observation. It was a science with a limited empirical basis, to be sure. A measure of that is that I could present a reasonably complete survey of the science (apart from the subtleties of the astronomical observations that the title was meant to indicate I would not attempt to address) in just 280 pages. I marshaled evidence for the homogeneity assumption - the cosmological principle - and concluded that the case was encouraging but not definitive. A decade later the case was much stronger, but resistance to the assumption died out more slowly, a not unusual phenomenon. The last section in the chapter on the Primeval Fireball - the name John Wheeler had suggested for the CMBR - has the title Is this the Primeval Fireball? My answer was cautious, largely because of the apparent anomaly in the measurements of the spectrum at wavelengths near 1 mm. The case for the fossil interpretation of the CMBR is close to compelling now: we have a vastly improved spectrum measurement, and detailed evidence that the radiation has the predicted disturbances caused by its interaction with the mass distribution at decoupling and along the line of sight. And we have the elegant concordance of the theory and observations of helium and deuterium. But all that is the subject of Chapter 5.
I close with some thoughts inspired by reading the exchange of letters between Dennis Sciama and Bob Dicke. In his reply, dated December 30, 1966, Bob writes
I was very happy to learn that you have abandoned steady state theory, but I do not recommend that you take too orthodox a position. A number of peculiar things are showing up that favor the scalar-tensor theory and I had hoped that you would be one of the few people who might be convinced when the observations were good enough to warrant it. Another reason for being unorthodox is that it's fun.
This is very characteristic of the central lesson Bob gave me by word and example: don't take received wisdom too seriously, but you had better take the science very seriously. Starting in Peebles (1984, 1986) I enjoyed better than a decade of fun pointing out the observational challenges to the then orthodox adoption of the Einstein-de Sitter cosmology (without Einstein's cosmological constant, and negligible space curvature). I spent about as much time constructing alternatives to the CDM model (with its set of assumptions about the initial conditions for structure formation). I meant this model (Peebles 1982) to serve as a simple example showing why the improving limits on the anisotropy of the CMBR were not necessarily inconsistent with the idea that galaxies and clusters of galaxies grew by gravity out of small primeval departures from homogeneity. As the model became popular I became nervous, because it was easy to think of alternatives that could equally well fit the still very loose observational constraints. Here again the consensus developed before the evidence warranted it, but in this case the orthodox view proved to be on the right track. I folded at Peebles (1999) when the advances in the tests had become so rapid that my alternatives were being ruled out as quickly as I could produce them.
The observational basis for cosmology now is far better than anything I would have imagined in the 1960s, and the case for the hot big bang far more compelling. With Dennis Sciama I have become "more orthodox than the orthodox." But with Bob Dicke I doubt that we now know all the physics relevant for the observational analysis of the evolution of the universe and its contents from high redshift (let us say from light element production, 2 ~ 1010) to the present. We are attempting to draw spectacularly large conclusions from what still is an exceedingly limited collection of data. I expect this active field of research will continue to fascinate the next few generations.
Was this article helpful?