Martin Harwit is Professor Emeritus of Astronomy at Cornell University and former Director of the National Air and Space Museum. He is a mission scientist on the European Space Agency's Far-Infrared Submillimeter Telescope project, Herschel, of which the National Aeronautics and Space Administration is also a sponsor.
In 1963 I initiated an effort to look for the cosmic background radiation from space (Harwit 1964). The small research groups I started, first at the NRL in Washington, DC, and later at Cornell University, designed and constructed cryogenically cooled rocket telescopes to detect this radiation. My calculations showed that we would be hindered by zodiacal foreground radiation. Our telescopes eventually confirmed this, by detecting the zodiacal glow along with a number of strong diffuse galactic sources. We also obtained painfully false results on the submillimeter component of the 3-K microwave background radiation, due largely to emission by contamination carried aloft by our rockets (Shivanandan, Houck and Harwit 1968). In order to depict the many mishaps, missteps, and misconceptions that motivated me to initiate background observations in the early 1960s, I begin my account ten years earlier, when I was a graduate student.
In the spring of 1954 I found myself standing in the Physics Department office of Professor David M. Dennison at the University of Michigan in Ann Arbor. Dennison was an eminent molecular theorist. Sitting behind his desk, he was finding it difficult to tell me that I would not qualify for a PhD in physics because I really had no aptitude for science. Perhaps I should look at other occupations, because science clearly was not my metier.
Two months earlier, I had turned 23, and my career as a scientist already was reaching an unfortunate conclusion. I had come to Michigan to study chemical physics. Although my undergraduate major had been physics, I had taken an advanced chemical physics course before coming to Michigan and read through Linus Pauling's (1948) excellent book The Nature of the Chemical Bond, and Gerhard Herzberg's (1945, 1950) two books on molecular spectroscopy. The field looked genuinely exciting. At Michigan, I was assigned to carry out near-infrared spectroscopic work on peptide bonds in the laboratories of Professor G. B. B. M. Sutherland, who later became Director of the National Physical Laboratories in Britain. During my one-year apprenticeship, I learned a lot about infrared techniques but did not accomplish much. I was studying for my doctoral exams at the time, and neither the research nor the exams went well, which was why I was standing in Professor Dennison's office that day.
Having to leave the Physics Department and wanting some time to figure out what to do next, I stayed on in Ann Arbor that summer, and found a job in the laboratories of Professor Leslie Jones in the School of Engineering. He and his group were conducting upper atmosphere research with rocket-borne instrumentation. I did some optical design with the group and tackled whatever jobs needed doing. It was the first time I had real fun in science.
The war in Korea had not yet ended in 1954. In the fall of that year I received the then standard letter from my draft board, which began with the ominous words, "Greetings from the President of the United States," and explained how my "friends and neighbors" had selected me to serve in the United States Army. I was to report for my two-year stint of duty in January 1955.
Because I had earned an MA in Physics at Michigan by this time, the Army assigned me to the Chemical Corps at the Army Chemical Center in Edgewood, Maryland. This is where I began my real scientific training. Most of the civil servants in the Corps were chemical engineers and knew little about physics. But now, ten years after the end of World War II, the government was asking them to work on radioactive fallout, neutron doses from nuclear bomb bursts, and similar problems. I found that with a few visits to the base library, I could usually figure out what needed to be done, and although I was just an army private, I was given a lot of responsibility. Nevertheless, my civil service supervisor would send me to places like MIT or Woods Hole to verify with known experts that my calculations had been correct, and I enjoyed the opportunities offered by those visits.
In my second year in the Army, I was sent to Eniwetok and Bikini atolls in the Pacific for a few months to participate in what at the time was believed to be the first hydrogen bomb drop from an aircraft. We attempted to measure neutron doses at different distances from nuclear explosions, big and small. Some of them could vaporize an entire island in the atoll, others just left a small crater. To while away the time between work and snorkeling in the waters of the atolls, I had taken along a number of books, among them a popular astronomy book by Fred Hoyle. I no longer recall whether it was his Frontiers in Astronomy or The Nature of the Universe. Both books were out in paperback by that time, as were all the books I had taken along. Though Hoyle used no formulae and little technical language, I began to think that I would be able to do the calculations he was describing. It was quite fascinating.
At the end of my two years' service, I applied to graduate schools and was accepted by MIT on the strengths of what must have been great recommendations from Leslie Jones, my supervisor at the Army Chemical Center, and one of the MIT professors the Army had sent me to consult.
At MIT, there was no course requirement in one's major subject. But for a minor, a student was required to pass three advanced courses. I signed up for an astrophysics minor. It is hard to believe, today, but in 1957 MIT had no astrophysics curriculum. However, an exchange arrangement with Harvard permitted me to take three graduate courses there.
Tommy Gold had just been given a Harvard professorship. At the time, he was postulating that dust on the Moon would hop around in response to electrostatic bombardment from the Solar wind. Inspired by this, I begged and borrowed some equipment in the Research Laboratory of Electronics at MIT, learned how to blow glass so I could construct a vacuum tube for bombarding dust with electrons, and then saw the dust disperse when I turned on the electron beam. Tommy came down to MIT to see this late one evening. He was delighted and asked whether I might like to switch to astrophysics after receiving my PhD. I had done some calculations in one of his seminars, and he thought I should postdoc with Fred Hoyle. Of course, I was very pleased, though I still had my thesis work to complete.
I had come across the recently discovered Hanbury Brown-Twiss effect, and read the controversy surrounding it that aired in the journal Nature at the time. Edward M. Purcell's clean resolution of that controversy was particularly illuminating. I thought that the techniques developed for detecting the HB-T effect might provide a first opportunity to directly detect Bose-Einstein fluctuations in electromagnetic radiation from a source in thermal equilibrium. None of the experimentalists in the MIT Physics Department was particularly interested in my making these measurements, but Professor William P. Allis, a leading plasma theorist, said he would be willing to supervise the thesis if I could find the means to build the requisite apparatus.
The Naval Supersonic Wind Tunnel located on the MIT campus at the time was run by Professor John R. Markham of the MIT Aeronautical Engineering Department. One of the problems they were tackling was the detection of the hot exhausts of rockets and jet engines. This necessitated devices sensitive to the infrared radiation from these plumes. Improved sensitivity could be achieved by using not one detector, but two, and correlating their signals. This correlation technique was also needed for the Hanbury Brown-Twiss apparatus. The Aeronautical Engineering Department offered to buy as much of the requisite equipment as could be commercially obtained. They would use the apparatus during working hours, and I was free to use it for my thesis work at night. They generously also provided me with the assistantship I would need to finish my thesis work.
The fluctuations to be measured were minuscule, and detectors available at the time were still quite insensitive, but by April 1960 I had reasonably reliable results, and was finished with my thesis (Harwit 1960). Early in May, my wife Marianne and I embarked on the USS United States for me to spend a NATO-sponsored postdoctoral fellowship year with Fred Hoyle in Cambridge. Four years earlier, I had been inspired by his popular writing. Now, I would be working with him. I hardly believed my good fortune!
When we arrived in Cambridge, Fred was away on one of his prolonged visits to Caltech, and I had time to finish a paper I had begun while still at MIT. I had found a small error in a paper on galaxy formation in a steady state universe by Dennis Sciama. When I redid the calculation, it showed quite clearly that there was no way that a steady state universe could form galaxies at the replenishment rate required by the expansion of the universe, unless forces other than gravitation were at play.
I submitted the paper to the Monthly Notices of the Royal Astronomical Society, and some time later received an acceptance and an invitation to present the work at one of the monthly meetings in Burlington House (Harwit 1961). To my dismay, a week before my scheduled talk, I saw an announcement on one of the Cavendish Laboratory's bulletin boards that Hermann Bondi, one of the original creators of the steady state theory, was going to give a talk at King's College, London, on precisely the same topic of steady state galaxy formation the week after my talk at the RAS. I had heard that Bondi was a fierce debater. As secretary of the RAS he would undoubtedly be present at my talk. I expected a punishing onslaught, and at once began to prepare myself by reading everything Bondi had ever written on related subjects.
On the day of the meeting, I gave my talk, sparred with Bondi, but felt that I had acquitted myself reasonably. At the end of the session, I approached Bondi and introduced myself. At his suggestion we went to eat a hamburger and chat for a while before he had to take the train home to Sussex and I returned to Cambridge. He told me he had been the referee on my paper which had suggested some further work to him. Would I have time to come to King's College the following week to hear his talk? I was delighted, of course.
Munching on our hamburgers that evening after my RAS talk, I remember us talking about the future. I mentioned that on my return to the United States, I hoped to set up equipment to carry out infrared astronomical observations. Nobody was active in that area, and yet it seemed highly promising for astrochemical studies with infrared spectrometers.
After a great year in Cambridge, working with Fred Hoyle after his return from Caltech and writing a few papers with him, I returned to the US, to take up an NSF postdoctoral fellowship, this time at Cornell University where Tommy Gold had invited me to come. He had just moved to Cornell to start a powerful new department.
After my fellowship year, I accepted a one-year assistant professorship at Cornell, at the end of which I was free to take a leave of absence. I knew I wanted to carry out infrared astronomical observations and felt that ultimately infrared spectroscopy would offer great insights. But the Earth's atmosphere absorbs much of the infrared spectrum and, even worse, glows strongly in the infrared. To obtain a clear view of the sky in this wavelength band, I knew I would need to take telescopes above the atmosphere; moreover, these telescopes would have to be cooled to cryogenic temperatures. Otherwise the glow from the telescope would be far stronger than any celestial signal. At MIT I had built a sensitive cooled infrared apparatus. At Michigan, in Sutherland's laboratory, I had gained experience with spectroscopy, and in Leslie Jones's group, I had learned how to build apparatus carried aloft in rockets. All I needed to do was to put all this together.
At Tommy Gold's suggestion, I visited Herbert Friedman of the US NRL in Washington, early in 1963, to propose the possibility of starting an infrared astronomy program using rocket-borne telescopes. NRL had impressive credentials in ultraviolet and X-ray observations from rockets, but had not ventured into the infrared.
Friedman was very receptive. In a friendly meeting held in his offices, we agreed that I would come to work at NRL in the fall of 1963 and stay for a year, with fellowship support from the NSF. During this year, I would set up a group of NRL scientists and engineers to conduct a program in rocket infrared astronomy. At the end of the year, I would return to Cornell to set up a similar program there, and the two research groups established in this way would thereafter continue to compete in the newborn field.
During the summer of 1963, I sought to clarify the steps we would take. It was clear from the start that we needed to keep our efforts simple; the telescopes would have to be small. Our first efforts would have to be broadband photometry; spectroscopy would have to be delayed until we had more experience with the far simpler photometry. But even with these limitations, we thought we should be able to obtain reasonable measurements of large-scale features and an isotropic background. For background observations, a small telescope would suffice as long as it had a high throughput, i.e., it maximized the product of telescope aperture and angular beam dimension on the sky.
The background radiation I hoped we would observe was radiation I thought should have been emitted in the conversion of hydrogen into helium over the eons. Even though I had written two papers, while in Cambridge, to show the difficulties the steady state theory had in accounting for galaxy formation, I still thought that all the helium now observed must have been produced in stars. Like most astrophysicists at the time, I was unaware of the pioneering work of Ralph Alpher and Robert Herman (1948). Unfortunately, most of it had been largely ignored, forgotten, or discounted.
In 1963 the helium content of the universe was known to account for approximately one quarter of all the atomic mass in the universe. If the conversion of hydrogen into helium had all taken place in stars, then some of the energy liberated in the process should be observable in the infrared. I no longer recall why I thought the observation was feasible, but this was an easy calculation, and there were so many things like that "in the air" at the time. There just weren't very many astrophysicists then interested in cosmo-logical questions, and many of these thoughts simply remained unpublished, though knowledgeable people were aware of them and exchanged ideas about them over tea or coffee. These were quick ideas that were not sufficiently substantive to warrant publication. They were somehow too obvious.
With thoughts about the accumulation of starlight in mind I presented a paper at a colloquium held at the University of Liege in late June 1963. In the proceedings of the conference I wrote (Harwit 1964)
(A)n interesting infrared observation concerns the frequently discussed suggestion that the overall cosmic background radiation might amount to as much as 3 x 10"11 watt/cm2 in the infrared ...
(T)he cosmic flux could only be detected from the immediate vicinity of the Earth, if the radiation were concentrated in a very long wavelength spectral range where interplanetary dust grains are expected to emit inefficiently.
I showed that the thermal emission of the zodiacal (interplanetary) dust cloud would dominate the brightness of the infrared sky in the near- and mid-infrared part of the spectrum and wrote,
One now is in a position to discuss the detrimental effects that zodiacal dust reradiation will have on infrared astronomical observations ... (T)he nature of the most promising infrared observations is different from much of the work in the visible region. One often hopes to obtain information about diffuse sources of radiation, so that the zodiacal foreground glow may be an important hindrance ... At 42 j this cloud would radiate of the order of 4 x 10 13 watt/cm2-sterad-^ at large elongation angles within the plane of the ecliptic.
Even today, four decades later, the zodiacal glow remains an obstacle to determining the true extragalactic background in the near- and mid-infrared. We may ultimately have to rely on teraelectron volt (TeV) observations of distant active galactic nuclei to determine the rate at which this gamma radiation is destroyed through electron-positron pair formation, as it transits through the cosmic infrared background in extragalactic space.
The Cornell-NRL collaboration started in earnest in September 1963. A large number of technical problems had to be overcome in just 12 months if the work of the first year was to culminate in demonstrable success. NRL provided major resources to the effort. Joining me were scientists Douglas McNutt, Kandiah Shivanandan, and Blair Zajac, mechanical engineer Henry C. Kondracki, and electronic engineer John M. Reece.
Though the ultimate goal of the group was to construct telescopes cooled to liquid helium temperatures which would offer unencumbered observations across the entire spectral range from 1 pm out to several hundred microns, we quickly realized that the design of a liquid nitrogen cooled telescope would be considerably more simple. Such a telescope, though not as cold, would still make possible near-infrared observations of great sensitivity, since a telescope cooled to the temperature of liquid nitrogen, ~80K, would emit negligible thermal radiation at short wavelengths, and the near-infrared detectors in any case should operate optimally at this temperature. Once sufficient experience in the construction of these near-infrared telescopes was gained, we intended to quickly turn to the technically more difficult task of constructing liquid-helium-cooled telescopes that could be operated at temperatures of 4 K with the helium at atmospheric pressure, or ^2 K if the helium was pumped down to very low pressure.
Many of the first launches were failures. Today, rocket launches have a better track record. But in the mid-1960s, failures of small sounding rockets to de-spin, pointing mechanisms to correctly orient the payload, delayed launches, and other problems often led to dismaying setbacks. Our efforts, like those of many others, were plagued by these difficulties.
I returned to Cornell University in the fall of 1964, whereupon Douglas McNutt took over the direction of the NRL group. We continued to collaborate on efforts that had been jointly started, but as these were completed, the two groups began to work independently and compete.
Shortly after my return to Ithaca, discussions with Dr. Nancy Roman, in charge of the astrophysics program at NASA resulted in grants to Cornell of an initial sum of $250,000 and annual budgets of $100,000, sufficient support to conduct a viable research program, initially with Aerobee 150, and later with the larger Aerobee 170 rockets. While NASA provided this initial outlay, we also obtained funding from the Air Force Cambridge Research Laboratories (AFCRL).
At Cornell I hired Henry C. Kondracki, who left NRL to move to Ithaca, New York, as full-time mechanical engineer. William Wernsing, an electrical engineer in Ithaca, also joined the group, as did Jim Dunston, a local jack-of-all-trades technician. James R. Houck, a graduate student just finishing a PhD at Cornell in solid state physics joined our small group after a couple of years. He was soon asked to join the Cornell faculty and the two of us established a long-lasting collaboration.
Constructing a liquid-helium-cooled telescope turned out to be a major engineering effort. A cryogenically cooled telescope had to be launched under vacuum. Otherwise, atmospheric gases would immediately condense on the optics. But vacuum vessels at that time tended to be constructed with thick steel walls making them far too massive to be launched on small rockets. A sufficiently light-weight design was needed. The thermal/mechanical design problem of constructing such a telescope, which could survive the vibrations and linear accelerations of launch and yet have minimal heat-conduction paths to the outer shell at room temperature, was difficult to solve.
Since the sensitivity of cryogenically cooled detectors in a cryogenically cooled telescope would be extremely high, observations were possible at high speeds. The bolometers favored by many ground-based observers were too slow to take advantage of this speed. Some of the photoconductors that had been developed for military purposes were far more promising. But it was soon apparent that the very low radiative background that a fully cooled telescope provided minimized the photon flux on these detectors, and correspondingly lowered the conductivity of the detector material. The detectors then attained extremely high resistances ranging up to 1011 Q. Even small capacitive effects would then produce unacceptably slow response times. A major effort had to be undertaken to decrease response times and take full advantage of the detectors' potential sensitivity and speed.
The Earth's surface brightness in the infrared was expected to be nine orders of magnitude higher than the basic signals the detectors were able to detect from their fields of view on the night sky. Extreme care had to be taken to baffle the telescope to eliminate any stray light from the Earth's limb that might be scattered or diffracted into the telescope.
It took us five years, and a succession of failures, before we were able to produce a successfully working liquid-helium-cooled astronomical telescope. Early designs incorporated a parabolic primary mirror with 18-cm aperture and focal ratio f/0.9. At altitude the entire telescope, except for the entrance aperture, was surrounded by liquid helium. We flew three different types of detectors on these flights - copper-doped germanium, gallium-doped germanium, and n-type indium-arsenide hot-electron bolometers - to cover progressively longer wavelengths between 5 ^m and 1.6 mm (Harwit, Houck and Fuhrmann 1969).
We had, of course, been aware of the Penzias and Wilson (1965a) discovery of the microwave background radiation. Its stunning cosmological implications were widely discussed. This was truly exciting work and we were eager to check it out. But, it was not until 1968-1969 that our liquid-helium-cooled telescopes began to reliably work and we were able to attempt the detection of the expected submillimeter component of a background flux at -3K.
Even with a well-working telescope, we encountered difficulties in background radiation measurements. These produced a false signal with all the characteristics of an isotropic flux at the longest wavelengths, 400 ^m to 1.3 mm. My colleagues and I initially reported these signals as possibly of cosmic origin (Shivanandan, Houck and Harwit 1968). But subsequent flights appeared to verify our findings (Pipher et al. 1971).
Part of the problem was due to rocket exhausts and other gaseous and particulate ejecta that accompanied the payload to great heights to form a diffuse, radiating cloud surrounding the telescope. More serious was diffracted radiation from the Earth's limb. Malcolm Savedoff of the University of Rochester first suggested to us that this was a potential problem. Although we had carefully checked to make sure that no significant amounts of scattered light could reach our detectors, we had no way of checking for diffracted submillimeter radiation. Our detectors worked only in a liquid-helium-cooled environment and we had no cryogenically cooled, evacuated test chamber to make the requisite tests. Jim Houck, however, redesigned the telescope's baffle structure for the next scheduled flight, and this effectively eliminated the false background signal we had observed.
The first successes of our rocket flights involved two quite different types of detections, and resulted from separate flights on December 2, 1970, and half a year later, on July 16, 1971. The first of these discovered and accurately measured the infrared radiation emitted by the circumsolar zodiacal dust cloud (Soifer, Houck and Harwit 1971). We detected radiation in three spectral ranges, at 5-6, 12-14 and 16-23 ^m. At 70-130 ^m we could initially place only an upper limit. More careful analysis provided a detection even at these long wavelengths (Pipher 1971). Both the three- and four-color photometry put the dust temperature at ~280K. My greatest surprise in these findings was that the dust radiated significantly more powerfully than I had predicted (Harwit 1964), indicating that the zodiacal dust grains were unexpectedly dark, scattering only a small fraction of the incident light, while absorbing and reemitting an appreciably larger portion. Thirty years later, I was pleased to see that the far more comprehensive COBE results of Kelsall et al. (1998) showed good agreement with the surface brightness of the zodiacal dust our rocket instrument had recorded.
The second discovery, made with the Cornell liquid-helium-cooled telescope on July 16, 1971, was the magnitude of the total infrared flux emanating from the galactic center and four other regions in central portions of the Milky Way at 5, 13, 20, and 100 ^m (Houck et al. 1971). The 85-115 ^m integrated flux over an area of 3° x 2° around the galactic center was 7 x 10"20 Wm_2 Hz_1, in excellent agreement with the balloon-borne result that had previously been obtained by Hoffmann, Frederick and Emery (1971). Excellent agreement for this wavelength range was also obtained for the galactic ionized hydrogen regions Messier 8 and NGC 6357. But the Cornell rocket flight also recorded the previously inaccessible flux from these three regions at 5-6, 12-14, and 16-23 ^m (Soifer, Pipher and Houck 1972). Additional results cited by the same authors from an earlier flight provided the 100 ^m flux for NGC 1499, a region previously unobserved at this wavelength.
More than a dozen years later, scans of the galactic center were also undertaken by the IRAS (Gautier et al. 1984). Though these authors did not compare their results to any previous work, the 100 ^m maps of the galactic center published by the IRAS team gave peak fluxes which, within normal calibration uncertainties, were essentially identical to the Cornell rocket results published a dozen years earlier. Within such uncertainties, the 12 ^m IRAS fluxes and the 12-14 ^m Cornell detections also showed reasonable agreement.
Even though our paper on the background measurements at 400 ^m to 1.3 mm was laced with cautionary comments, it gathered widespread attention. As we became convinced that the signals were actually due to contamination we withdrew the results but, for the next 30 years, I continued to feel badly about this mistake. Not until Jean-Loup Puget and his group derived the far-infrared flux from COBE scans did I begin to feel relieved (Puget et al. 1996). If the correct analysis of the various cosmic background components had taken more than another quarter of a century and ^$500 million, roughly five hundred times more money than we had spent in the course of our entire rocket program, it was perhaps not so shameful to have been wrong. Sometimes it may be better to try difficult observations and fail, than not to try at all.
Was this article helpful?