Pre-1949: When all radio sources are Galactic stars (1:59 – 4:20)

I’d like to talk about the events leading up to the discovery of quasars (and add a few remarks after that) as seen from the eyes of a radio astronomer when the discrete radio sources were first being discovered in the mid-1940s. They were widely believed to be stellar and were called radio stars. This is natural. Carl Jansky and Grote Reber had discovered radio emissions from the Milky Way. The Milky Way is composed of stars; therefore, discrete radio sources were probably stars.

This moniker should have changed as early as 1949. John Bolton, Gordon Stanley and Bruce Slee measured the accurate position of three strong radio sources.  At that time position accuracies were a degree or so. They were using the Cliff Interferometer, and they measured the position of three strong radio sources, which they identified with NGC 5128, M87, and the Crab Nebula. Their 1949 paper mostly discussed the Crab Nebula, which was a well-known object to astronomers at the time. At the very end of their paper, there is one paragraph where they discussed NGC 5128 and M87 and they wrote “NGC 5128 and NGC 4487 (M87) have not been resolved into stars, so there is little direct evidence that they are true galaxies. If the identification of the radio sources are accepted, it would indicate that they are [within our own Galaxy.]” In fact, the title of the paper is “Positions of Radio Sources of Galactic Radio-Frequency Radiation.” Nevertheless, history has treated them well and this paper is given credit for the discovery of radio galaxies. Years later, I asked John Bolton about their conclusions, since it seemed obvious, even at the time, that these objects were extragalactic. So I asked John, “How could you have said they were Galactic?” And he said something to the effect, “Well, I knew that if they were actually extragalactic the radio luminosity would be fantastic and I was afraid the referee would give us trouble. So we just said they were Galactic.” We're going to come back to this [bet hedging] because it happens over and over again.

Radio galaxies (4:27-6:34)

Things were clarified in 1954 when Walter Baade and Rudolf Minkowski identified Cygnus A, which is a radio galaxy with the redshift of z=0.06. This is the strongest radio source in the sky and is relatively bright, 17th magnitude. It was immediately realized by everybody that this meant that the weaker radio sources would probably be galaxies at very much larger distances, and therefore higher redshifts, perhaps at the limits of the 200-inch telescope or even beyond. The race was on to find radio galaxies using radio sources as an indicator of very large redshift galaxies.  Incidentally, Cygnus A was actually identified [as a galaxy] three years earlier by Bernie Mills in Australia, well before the Baade and Minkowski paper. Minkowski acknowledged that he did not accept the Mills identification because he couldn't believe that such a strong radio source could come from such a faint galaxy. So that's the second time that rejection of a radio source as a distant galaxy happened.

Now, Australian John Bolton came to Caltech in 1955 with the expressed purpose of building a radio telescope to find faint radio sources and measure their accurate positions in order to find distant galaxies. He was immediately successful. It went into operation in 1960 and yielded positional accuracies on the order of five or ten arcseconds. This led to the identification of the 20th magnitude galaxy 3c 295, which is a double source just like Cygnus A. It’s about 10 times smaller and is 10 times more distant (Minkowski measured z=0.46), so everything just fit. The new paradigm at the times was that all the discrete radio sources at high latitudes (out of the plane of the Galaxy) were galaxies, and they were called radio galaxies. 

3C 48: The first radio star (6:40-10:27)

The push to find more distant radio galaxies was feverish. Concentrating on small diameter radio sources led to the discovery of 3c 48 later on in the year by Tom Matthews and John Bolton using the Caltech interferometer. Jesse Greenstein, Guido Munch, and Allan Sandage all took spectra. They found emission lines which they didn't understand. Allan Sandage was at the AAS meeting in New York in December of 1960. At the last minute, I think just because he was there, he was asked to give a late paper on the identification of the first radio star. Unfortunately, at that time, the abstracts of late papers were not published in the Astronomical Journal. I asked the AAS office if they had these abstracts on file. They said everything has been sent to the American Institute of Physics, but somebody from the AAS staff volunteered to go over the AIP and look for this abstract. She wrote back to me that, well, she found all the records of all AAS meetings before this one and all the ones afterward this one, but the records of the 107th Meeting were just missing. So, the only written record of what actually happened at that AAS meeting is a few sentences in the Carnegie annual summary and also a short article or newsletter published a few months later in Sky and Telescope written by the magazine's editor. Both accounts stated something to the effect that there's a remote possibility that 3c 48 may be a distant galaxy of stars, but there's a general agreement that it's a relatively nearby star.

Eds. Note: this may be why Sandage is often given sole credit, or the only collaboratoer mentioned, for the discovery of 3c 48 in so many accounts of the discovery of 3c 48, because he was the sole individual within the Matthews, Bolton, Greenstein, Münch and Sandage collaboration who presented the discovery at the AAS meeting.

The evidence for 3c 48 to be a Galactic star was overwhelming as it was unresolved both in the radio and the optical (less than an arcsecond), whereas most radio galaxies were several or tens of arcseconds. Even though it had a very peculiar spectrum, which nobody understood-- lots of emission and absorption lines and a continuum that was very blue-- a critical thing was that it was variable. Only standard [stars] had measured variability of, I think, nearly half a magnitude over a few months, and it certainly seemed convincing that 3c 48 was also a star. Jesse Greenstein wrote a very exhaustive paper with the title “The radio star 3c 48”, in which he interpreted 3c 48 as the stellar remains of a supernova. He discussed the spectrum, which he interpreted as highly ionized states of various rare earth elements. The typewritten manuscript is about 40 pages long and had about 100 equations.  Buried, I think about page 10, is one short paragraph where he says that except for z=0.367, no redshift explains the strongest lines of any single ionization, but that the case for a large redshift is definitely not proven and 3c 48 is therefore not an extragalactic nebulae.  So instead of emphasizing the [high redshift] he did the reverse and downplayed it; he said it's not proven. In the next months, two more so-called radio stars were discovered, 3c 196 and 3c 286.  Matthews and Sandage published a paper [titled “Optical Identification of 3c 48, 3c 196, and 3c 286 with Stellar Objects”] on these three sources and they also said there's no plausible combination of redshifted emission lines. [Note “Identified ... with Stellar Objects” in the title.]

3C 273 as a distant galaxy (10:33-12:50)

Now for the famous story of the lunar occultation of 3c 273-- there were actually three occultations in May, August, and October. Following the August occultation John  Bolton wrote a letter to Maarten Schmidt. The letter was mostly about their ongoing program of radio galaxy identifications. But in that letter, Bolton tasks Martin to get a spectrum of an identified radio source at minus 28 degrees because he wanted to use that as a calibrator for his Parks Radio Telescope occultation program. At the very end of the letter, he provides what the positions of  3c 273 are from the occultation measurements. He gives the positions of component A and component B. That was in August 1962. 3c 273 was too close to the Sun at the time and Maarten had to wait until December before trying to get a spectrum.  As it turns out, the positions that Bolton sent in August were in error by about 15 arcseconds; it wasn't until January of 1963 that Bolton sent a corrected position. Cyril Hazard had made an arithmetic error in reducing the occultation data. So, it wasn't until January that Martin had the correct position. 

However, by that time he had taken the spectrum already. 3c 273 is one of the few sources, maybe the only one in the sky, where the radio and optical morphologies are very similar. There was an optical image at the time, and the jet was thought by the experts, including Minkowski, to be an edge on galaxy. I think maybe Maarten suspected that was the case too. 

In December 1962, Maarten Schmidt (1929-2022) had the positions he received in October (from Bolton's August letter). In order to eliminate the brightest nearby object, he obtained the spectrum of the 13th magnitude star closest to the the erroneous position of component A. His spectrum showed the classic blue intensity ramp up with broad emission lines seen for radio stars, so he was convinced he had found the optical counterpart to the radio source 3c 273.) (left) March 1963, Marteen Schmidt decodes the spectrum of 3C 273.  He measured a redshift of z=0.153. It instantly became the highest known redshifted object. (right) The spectrum of component A of 3C 273 showing its redshift Balmer lines and its increasing blueward flux.

Greenstein’s preprint and the CSIRO / University of Sydney controversy (14:25-16:56)

In early 1963, there was a quick flurry of publications in Nature. First, was the occultation paper by Cyril Hazard, Brian Mackey, and John Shimmins. Mackay and Simmins were from Australia’s Commonwealth Scientific and Industrial Research Organization (CSIRO) and were assigned to work with Hazard, who was affiliated with the University of Sydney; Mackay and Shimmins were both engineers assigned to Parks Observatory. Second was Maarten's paper on 3c 273 identified at z=0.15. Third was Bev Oke’s identifying the H-alpha line in the spectrum of 3c 273 at z=0.15. Jesse Greenstein, as soon as he and Maarten realized via follow-up inspection that the 3c 48 redshift is z=0.37, apparently forgot that he had mentioned this redshift in his paper [the one in which he said 3c 48 was a radio loud supernova remnant]. Greenstein had passed out preprints to lots of people in the department at Caltech and he soon withdrew the paper, which had already been accepted by the Astrophysical Journal.  Very quickly, he withdrew it and he went around collecting all the preprints. I had gotten my copy second hand from Tom Matthews and so he didn't know I had one and I kept it (and I think Maarten also has an “illegal” copy). Matthews and Sandage had prepared their paper (mentioned above), which they submitted in December 1962, before the redshift determination of 3c 48 was accepted. Recall, they had interpreted  3c 48, 3c 196 and 3c 286 as stellar objects.  But after the January or February realization of the large redshift, they added a note in proof that 3c 48 was a galaxy (see above).  

Maarten had mentioned [shady] business about the affiliation of Cyril Hazard in their occultation paper as it was being prepared to be submitted to Nature.  The draft has Hazard correctly affiliated with the University of Sydney and Mackay and Shimmins with CSIRO. But, before the paper was submitted by the CSIRO Publications Office, somebody in that office wrote the annotation "delete" on the original next to Hazard's affiliation. This caused a big international controversy! CSIRO of course claimed it was a mistake and happened because when the paper actually came out it recorded only the Chairman's CSIRO affiliation. They claimed it was an error in changing the paper from a letter format to an article format, but in fact I think the [written] evidence is pretty damming [that CSIRO underhandedly "grabbed the brass ring!"].

3C 48 Revisited, Bolton vs. Greenstein (17:02-20:54)

Just before Allan Sandage’s AAS paper, John Bolton left Caltech to return to Australia and supervise the completion of the Parks Radio Telescope. In 1989, he gave a talk at the Australian Astronomical Society, and he said the best fit he could find for the broad and narrow lines of 3c 48 were z=0.37. These are the lines Greenstein had mentioned in his retracted preprint. John is a radio astronomer; not an optical spectroscopist. But he said he had measured the redshifts based on these two lines to be z=0.37.  Well, John died a few years later and I wrote a memorial article for PASP in which I quoted him. I said back in 1960, two years before the 3c 273 business, that John Bolton had understood the redshift of 3c 48 to be z=0.37. I sent a draft to the manuscript to various colleagues, including at Caltech. Tom Matthews never responded. Maarten made a few constructive suggestions to the paper. But Jesse Greenstein went non-linear and he accused Bolton of a fabrication a lie and trying to take credit and so on and so forth. Now what am I going to do?! John Bolton was my thesis advisor-- everything I know about radio astronomy I learned from him. I greatly admired him. But he wasn't a spectroscopist and Greenstein was. Finally, I changed it slightly and instead of saying John Bolton had discovered the redshift back in 1960, I said John Bolton claimed he discovered the redshift in 1960, which was true.

Just a few years ago (circa 2010), a letter form John Bolton turns up in the National Library of Australia. It was dated November 16, 1960, about a month before the AAS meeting at which Sandage presented 3c 48 and before John left to go to Australia. In John's very characteristic handwriting, he writes it [3c 48] is not a star, that measurements in a high dispersion spectrum suggest lines of neon and argon and the redshift is z=0.367.  This is just the redshift that was in Greenstein's paper that he subsequently withdrew. John also realized the absolute magnitude is -24, two magnitudes greater than anything that was known. Apparently, between middle of November and when he sailed to Australia on December 12th, he talked further with Greenstein and Ira Bowen, both spectroscopy experts, and they convinced him that the three or four angstrom discrepancy between the lines in the spectrum was too great to be explained by observational error. When John’s boat stopped in Hawaii, he writes another letter and says 3c 48 is most likely a star. Some years later, I had lunch with Jesse Greenstein after he retired and I asked him, “how could you say it has the redshift of 0.37, but then you went and described it as not proven and so on ?” He said that he had a reputation for being a radical and was afraid to go out on a limb with such an extreme idea. This is essentially the same argument that John had 25 years earlier about M87 and that Minkowski had about the about Cygnus A.

3C 48: Sandage’s blue stars and rebuttals (21:00-32:46)

As Martin described after the  3c 48 and 3c 273 identifications, the incentive was on to find larger and larger redshifts identifying more quasars. It was quickly realized that the quasar colors were very blue and you could more simply identify quasars without having precise positions if you just look for a blue stellar object close to the radio source. Allan Sandage went even a step further and realized that there are lots of blue stellar objects that had no association with radio sources and he wrote this paper “A Major New Constituent of the Universe: The Quasi-Stellar Galaxies” in which he argued that these quasi-stellar galaxies, or blue stellar objects, were a thousand times more numerous than the 3c radio sources. The interesting thing is the paper was received in the Astrophysical Journal office on May 15, 1965. The editor was apparently so impressed he didn't send it out to be refereed; instead he held up publication of the journal so it could appear in the May 15th issue! That did not go over well in the community! I think people felt Allan was given preferential treatment.

Everybody jumped on this. Tom Kendon and Roger Lynds pointed out that most of Sandage’s blue stellar objects were in fact Galactic. And Fritz Wicky of course jumped in, and like everything else, said he had discovered these long ago but called them compact galaxies. As we now know, the so-called radio quiet quasar (not radio silent, they are weak radio sources) are about 10 times more numerous than the radio loud quasars. It's been somewhat controversial over the years as to whether the radio quiet and the radio loud quasars are just part of a continuous distribution, such that the radio quiet quasars are just a low luminosity end of a continuous luminosity function, or whether they're a separate category of a phenomena.

Why 3C 273 was first to be accepted? (23:52-25:57)

The lore is that Maarten Schmidt observed 3c 273 because it was occulted. The truth is, the Parks Radio Telescope did observe it, but not as part of the quest to find high redshift objects. Tom Bolton and Cyril Hazard were intrigued by the occultation technique, so they took advantage of the opportunity. But, in fact, I believe that the so-called accurate position of the occultation actually played no role in the identification, other than to motivate Maarten to look in that general area of the sky. As he just pointed out in his talk this morning, and stated in the first line of his paper,  "the star is near the radio source;" he did not say coincident with the radio source. This is because at that time it wasn't coincidence (because the position he had was erroneous). Maarten didn't know that for another month! His ability to identify the curiously so-called large redshift the 3c 273 was only possible because it was small!  That is, the lines were shifted into the familiar optical part of the spectrum. That's why they couldn't identify 3c 48 earlier, because they’re all unfamiliar lines in 3c 48. The other radio stars actually should have been important, but in the end actually played no role in the discovery of quasars; it was only after 3c 273 and then 3c 48 that rehshifts very quickly increased up to the order of z=2.  

Final ruminations (26:06-31:11)

Even though astronomers were motivated by a classical cosmology to determine the Hubble-Lemaître constant and deceleration parameter, in fact quasars have played, I think, no real role in classical cosmology.  That's because they're not standard candles in either the radio or optical. They have a huge range of luminosity. That was used as one argument against the cosmological redshifts-- that there was no tight  luminosity-redshift relation. But quasars have had a big impact on astrophysics; we now recognize the existence of supermassive black holes from in the nuclei of galaxies and the role that supermassive black holes play in the formation and evolution of galaxies. I think quasars have had a big impact to the sociology of astronomy and also astronomers. Quasars led to proliferation of conferences such as this one. They also exacerbated existing tensions at the time between the Caltech astronomers and the Santa Barbara Street astronomers already mentioned and the whole culture of the non-cosmological redshifts by some very respected colleagues, or at least very respected up until this time.

Eds. Note: Arguably, the study of intervening quasars absorption lines has been and continues to be a huge impact on the progress of astronomy. This includes (1) the Lyman-alpha forest, large-scale strucutres, and evolution of the IGM, (2) D/H measurements and constraints on early-universe and Big Bang Nuclesynthesis and early-universe physics, (3) reionization physics and the impact of the first stars and galaxies and evolution of escape fractions and the background radiation, (4) the CGM and baryon cycle key to understanding galaxy formation and evolution, stellar feedback, and (5) the measurement of the cosmic evolution of metals and dust, in particular using DLAs.  

There are still a lot of questions in my mind. Why did it take so long to identify 3c 273? It's a 13th magnitude object! It's one of the strongest radios sources; I think the fifth(*) strongest source in the 3C catalog. The Owens Valley interferometer had positions good to five or ten arcseconds, but somehow that position never got from the first floor to the second floor of Robinson Hall at Caltech. Maybe it did, but with a typographical error or some other error. In fact Maarten had taken a spectrum of a galaxy which he thought was identified with 3c 273 back in May 1962, which turned out to be a minute of arc away. How did that error creep in? Why was the redshift of 3c 48 not accepted back when it’s position was discovered in 1960-61, two years before 3c 273 was discovered? As Maarten stated, it was just unacceptable; it was so luminous, so small, astronomers were just too conservative. But why was a three to four angstrom discrepancy in the lines enough for people to not believe the spectrum? Finally, (after all that happening before 1963), why did it take Maarten six full weeks to recognize it was simply a series of redshifted hydrogen Balmer lines in the spectrum of 3c 273?

* it's the 7th brightest

Like I said, I was a graduate student at the time. I was working on my own thesis. I was sort of aware of what was going on. I do remember somebody asked about the name quasi-stellar radio source and where that came from. There were a number of names that were floating around the department. Just a few days prior to the Nature letter being submitted, I remember Maarten came into coffee, and said, “Well, I'm sending it off today and I'm going to call it a quasi-stellar radio source unless somebody else has a better idea.”  And that was the default, but over the years I've had a good chance to talk to most of the people involved and they've been very generous in sharing their recollections-- particularly Maarten, who I keep bothering with those phone calls and emails. I very much appreciate all their help. 

Thank you.

There has been no AI assistance in the creation of the material on this page. 

In 2014, the year after this talk was delivered at Caltech, Dr. Kellerman published an article entitled "The Discovery of Quasars and It's Aftermath," published in the 2014 Journal of Astronomical History and Heritage, 17(3), 267–282' arXiv:1304.3627. In 2025, Dr. Kellerman delivered the 59th Karl Jansky Lecture entitled "Discovering the Radio Universe" (it is published as Kellerman, K. I. 2025, Journal of Astronomical Heritage, 28(3), 553-567; arXiv:2511.16779).  In these articles, he chronicles many of the historical anecdotes described here, as well as others.  They are enjoyable and informative reads. If you enjoyed his narrative here, then they are highly recommended. Dr. Kellerman, in collaboration with Ellen Bouton, has also published the book Star Noise: Discovering the Radio Universe, which provides a detailed history of radio astronomy, including an in-depth account of the discovery of high-redshift radio galaxies and quasars. This book is also a fun and informative read and is highly recommended.