Fig. 4.20. Record of our first 7-cm observation.
That was a direct confrontation. We expected 2.3 K from the sky and 1K from the absorption in the walls of the antenna, and we saw something that was obviously considerably more than that. It was really a qualitative difference rather than just quantitative because the antenna was hotter than the helium reference and it should have been colder. But we knew that the problem was either in the antenna or beyond. Arno's initial reaction was "Well, I made a pretty good cold load!" The most likely problem in such an experiment is that you do not understand all the sources of extra radiation in your reference noise source, but it is not possible to make it have a lower temperature than the liquid helium.
It initially looked like we could not do the galactic halo experiment, but at that time our measurements of the gain of the antenna had started (Hogg and Wilson 1965) and we wanted to go on with the absolute flux measurements before taking anything apart or trying to change anything. We ended up waiting for almost nine months before doing anything about our antenna temperature problem; however, we were thinking about it all that time.
We thought of several possible explanations of the excess antenna temperature. Many radio astronomers at the time thought the centimeter-wave atmospheric radiation was about twice what we were saying. That would have gone a long way toward explaining our problem. However, the curve for the zenith angle dependence in Figure 4.18 indicates that we were measuring the atmospheric absorption and emission correctly. It turned out later that the centimeter astronomers had applied refraction corrections to their measurements of radio sources in the wrong sense. John Shakeshaft finally straightened this out (Howell and Shakeshaft 1967a).
Since Crawford Hill overlooks New York City, perhaps man-made interference was causing trouble. Therefore, we turned our antenna down and scanned around the horizon. We found a little bit of superthermal radiation; but, given the horn-reflector's rejection of back radiation, nothing that would explain the sort of thing that we were seeing.
Could it be the Milky Way? Not according to extrapolations from low frequencies. The galactic poles should have a very small brightness at 7-cm and our actual measurements of the plane of the Milky Way did fit very well with the extrapolations.
Perhaps it was a large number of background discrete sources. The strongest discrete source we could see was Cas A and it had an antenna temperature of 7 K. Point sources extrapolate in frequency in about the same way as the radiation from the Galaxy, so they seemed a very unlikely explanation.
That left radiation from the walls of the antenna itself. We calculated nine-tenths of a kelvin for that. We took into account the actual construction of the transition between the tapering horn and the circular waveguide of the radiometer, which is the most important part. It was made of electroformed copper and we measured waveguides of the same material in the lab to determine the loss under real rather than theoretical conditions.
We had to wait some time to finish the Cas A flux measurement, but in the spring of 1965, almost a year later, we had completed it (Penzias and Wilson 1965b). The Earth had almost made a complete cycle around the Sun and nothing had changed in what we were measuring. We pointed to many different parts of the sky, and unless we had a known source or the plane of the Galaxy in our beam, we had never seen anything other than the usual antenna temperature. In 1962 there had been a high-altitude nuclear explosion over the Pacific which had greatly increased the amount of plasma in the van Allen belts around the Earth. We were initially worried that something strange was going on there, but after a year, the population of the van Allen belts had gone down considerably and we had not seen any change.
There was a pair of pigeons living in the antenna at the time, and they had deposited their white droppings in the part of the horn where they roosted. So we cleaned up the antenna, caught the pigeons in a havahart trap, and put some aluminum tape over the joints between the separate pieces of aluminum that made it up. All of this made only a minor improvement.
We were really scratching our heads about what to do until one day Arno happened to be talking to Bernie Burke about other matters. After they had finished talking about what Arno had phoned him for, Bernie asked Arno about our Galactic halo experiment. Arno told him about our dilemma of excess noise from the horn-reflector, that the Galactic halo experiment would not work, and that we could not understand what was going on. Bernie had heard from his friend Ken Turner about Jim Peebles' recent colloquium at Johns Hopkins where he described calculations of microwave radiation from a hot big bang. Bernie suggested that we get in touch with Dicke's group at Princeton. So of course Arno called Dicke. Dicke was thinking about oscillating big bangs which he concluded should be hot. After a discussion on the phone, they sent us a preprint and agreed to come for a visit. When they came and saw our equipment they agreed that what we had done was probably correct. Afterward our two groups wrote separate letters to The Astrophysical Journal (Dicke et al. 1965; Penzias and Wilson 1965a).
We made one last check before actually sending off our letter for publication. We took a signal generator, attached it to a small horn, and took it around the top of Crawford Hill to artificially increase the temperature of the ground and measure the back lobe level of the 20-ft horn; maybe there was something wrong with it. But the result was as low as we expected. So we sent the letter in!
Arno and I were very happy to have any sort of an answer to our dilemma. Any reasonable explanation would probably have made us happy. In fact, I do not think either of us took the cosmology very seriously at first. I had come from Caltech and had been there during many of Fred Hoyle's visits. Philosophically, I liked the steady state cosmology. So I thought that we should report our result as a simple measurement; after all the measurement might be correct even if the cosmology turned out not to be!
The submission date on our paper was May 13, 1965, and a few days later on May 21 my father was visiting us as part of a business trip. As was typical for him he woke up before I did and went for a walk. He came back with a copy of The New York Times which had a picture of the 20-ft horn-reflector and an article by Walter Sullivan entitled "Signals Imply a 'Big Bang' Universe" on the front page (Sullivan 1965). Besides being a very satisfying experience, this awakened me to the fact that the world was taking the cosmology seriously.
At the time of our paper, the spectrum of the CMBR was only determined by our measurement and an upper limit at 404 MHz which was dominated by galactic radiation. This was only enough to rule out ordinary radio sources. Soon after our result became known, George Field (Field, Herbig and Hitchcock 1966), Pat Thaddeus (Thaddeus and Clauser 1966) and Iosif Shklovsky (Shklovsky 1966) independently realized that the absorption of stellar optical radiation by CN in interstellar clouds, which had been known since 1940, could be used to measure the radiation temperature in those clouds. The measurements of those three groups indicated about 3 K for the radiation temperature at 2.6-mm wavelength. The Princeton group (Roll and Wilkinson 1966) completed their first measurement at 3.2 cm by the end of the year. Arno and I repeated our 7.35-cm measurement with a smaller horn-reflector with consistent results. We then installed a 21-cm receiver on the 20-ft horn-reflector and made a measurement (Penzias and Wilson 1967), which was consistent with Howell and Shakeshaft's 21-cm measurement (Howell and Shakeshaft 1966) made about the same time. In approximately a year there were seven measurements consistent with a 3-K CMBR, but it would be more than a decade before the spectrum was proven to be a blackbody spectrum rather than graybody, and thus definitively from the early universe. The details of these early measurements are covered by other essays in this chapter.
Looking back, it is a bit surprising how quickly our results were accepted among the astronomers I talked to. It probably helped that the steady state theory was failing to fit observations and Bell Labs had a reputation for doing good science. There were only a couple of occasions where I was challenged about the correctness of our measurements. More often, paradigm changes of that magnitude are resisted much more by established scientists.
It is interesting to compare the equipment Arno and I used to that which Roll and Wilkinson designed for the purpose of detecting the CMBR. Theirs had a large amount of symmetry between the path to the sky and that to the helium reference source, just the sort of thing a physicist would design. Ours required very careful measurement of the loss in the separate paths for making the comparison, but the high-sensitivity receiver and high-gain antenna had advantages in measuring the radiation from the Earth's atmosphere and in looking for and rejecting interference and foreground radio sources. We could make a measurement with a tenth of a kelvin accuracy in a second whereas they had to integrate a long time for that accuracy.
The ability to make meaningful tests in a short time can be invaluable when working with equipment which is not doing what you expect. In short, I think that our equipment inherited from other Bell Labs projects was ideal for finding something unexpected, but similar to what we were looking for, and theirs was more suited to a high-accuracy measurement. With hindsight, we should have explored the degree and larger scale isotropy of the CMBR more carefully before moving to 21 cm. Analyzing the records made for flux measurements of a number of sources on one day we were able to put a limit of 0.1 K on the large-scale anisotropy (Wilson and Penzias 1967). We could have made a measurement on a one degree to tens of degrees in angular scale, which would have been the most accurate for several years.
In 1966 Roy Tillotson, who had succeeded Rudi Kompfner as the director of our laboratory (an organizational unit of several departments at Bell Labs), told us two things which I still remember. First, he told us to identify and preserve our first record which showed the CMBR. Second, he reminded us that we had agreed (probably as a result of Congress having created COMSAT and taken the international satellite business away from AT&T) to each spend half time on radio astronomy and do things for the Bell System in the other half. Therefore, since we had been doing astronomy almost full time for several years, we should make good on the second part of the bargain. Over the next several years Arno and I continued to do 21-cm measurement with the 20-ft horn-reflector, but we were also involved in projects more directly targeted to communications.
For the first such project, Arno and I set up a propagation measurement at 10.6-^m wavelength between Crawford Hill and the Holmdel building a couple of miles away using one of Kumar Patel's first high-power CO2 lasers. It was hoped that one could communicate over short distances in the far infrared much more readily than in the optical and near infrared. Dave Hogg had shown that those wavelengths were highly attenuated over the same path in foggy weather. Alas, 10 ^m was much better, but not nearly good enough to be practical. It was, however, fun to convert the parts we got into a reliably operating laser.
I set up a small radio telescope to automatically track the Sun every day and measure its brightness as a way to explore the possibility of using bands at 1-cm and 2-cm wavelengths for domestic satellite communications. I showed that those bands were useful except during very heavy rains. I also found that if one were willing to have two Earth stations 5 or 10 miles apart one could work around the heaviest rain cells. I did this using fixed pointed radiometers which measured the radiation of Earth's atmosphere from which I calculated the attenuation. A somewhat longer wavelength band is currently used for direct broadcast satellite TV.
I was having considerable success and fun with the millimeter-wave propagation experiments and was drifting toward working more of the time on them, but we also continued our 21-cm work, especially with Pierre Encrenaz, a Princeton graduate student at that time.
Then in 1968 Arno suggested using a millimeter-wave receiver based on Schottky barrier diodes with NRAO's recently completed 36-ft antenna on Kitt Peak. Charlie Burrus, who was just down the hall from us, had developed the diodes and mixer assembly for a millimeter-wave (pre-optical fiber days) broadband communications system. This initial experiment demonstrated the feasibility of this effort, but produced little in the way of new science. We left that 90-GHz receiver for NRAO to use in developing the antenna. Two years later Sandy Weinreb of NRAO offered to provide a spectrometer and frequency control equipment for the 36 ft. We returned with a higher frequency Burrus receiver. Arno talked Keith Jefferts (a Bell Labs atomic physicist interested in millimeter-wave spectroscopy) and me into integrating the Bell Labs receiver into an NRAO receiver box that would fit at the focus of the 36-ft antenna. We would then go back to look for carbon monoxide in interstellar space. At one point in this process, Keith remarked that Arno had the two best technicians at Bell Labs wiring the receiver for him.
The payback came when Keith and I joined Sandy at Kitt Peak to get it all working. After several frustrating days, Sandy had to leave, but the next day we got it all tenuously working and put it on the antenna. I asked the telescope operator to point to the BN/KL region of the Orion Nebula where two nebulae which are bright, one in the near infrared and the other in the far infrared, would be in our beam. I was watching the rather crude output of the spectrometer when some of the center channels increased from their somewhat random previous outputs. The operator confirmed that we had just reached the source. I asked him to go off the source and the channels went back down. Thus in a few seconds, using a system which was hundreds of times less sensitive than the one on the 20-ft horn-reflector, we discovered carbon monoxide in an interstellar cloud. I had picked the BN/KL source because it was the source in our list of candidates which was overhead at the time, but it turned out that it is the strongest CO source in the sky. Arno arrived the next day to find that the key discovery had been made (Wilson, Jefferts and Penzias 1970).
The carbon monoxide and other simple molecules that we and others have found since can be thought of as stains which allow us to measure the structure and dynamics of the interstellar molecular clouds. The clouds are so cold that their main constituent, hydrogen, doesn't radiate. The radiation from simple molecules has shown that these dense molecular clouds exist, star formation is active in them and they are common in galaxies. Since that time, a large number of astronomers have worked on understanding the physical and chemical conditions in these clouds and the formation of stars within them. For several years after the discovery, Bell Labs gave Burrus diodes to other observatories and taught other groups how to make them.
This discovery changed the direction of my career. We spent five exhilarating years exploring interstellar clouds and discovering new molecules and their isotopic variants with our receivers and the 36-ft antenna at Kitt Peak.
I then became project director for the 7-m antenna. It was designed to do millimeter-wave astronomy when the weather was good and satellite propagation measurements at 1-cm and 2-cm wavelengths in weather bad enough to affect that band. We then had almost two decades of additional studies of molecular clouds and the cores around young stars which are embedded in them. The Crawford Hill astronomy group grew to include several additional people at its peak. Later the astronomy effort became less relevant to AT&T's need to prosper in the post-divestiture days and therefore declined. The Sub-Millimeter Array which I am working on now is an aperture synthesis array that spends most of its time observing radiation from the simple molecules and dust in these star-forming regions.
This work has taken me much closer to the origin of Earth and perhaps the organic molecules from which life originated, as opposed to the universe. On that larger scale, however, I have found the beautiful spectrum of the CMBR measured by COBE, and the evolving page full of accurate numbers derived from its fluctuations, immensely satisfying.
Was this article helpful?