1

Rayleigh-Jeans extrapolation

2.7 K Blackbody

%

Temperature

3.20 cm

2.69+ 16 K

1.58 cm

2.78+ 12 K

0.86 cm

2.56+ 17 K

0.33 cm

2.46+ K

Wavelength (cm)

Fig. 4.36. A 2.7-K brightness spectrum compared to the extrapolated Rayleigh-Jeans power law (determined by the A = 7.35-cm Bell Labs results and 3.20-cm point shown here) and superposed on White Mountain and Climax measurements. The ^30% departure of the Planck curve from the power law at A = 3.3 mm is easily visualized, but not so easily measured. As discussed in the text, lines of constant antenna temperature lie parallel to the Rayleigh-Jeans power law in this plot (adapted from Partridge 1969; ©1969 American Scientist).

isotropy (which Wilkinson and Partridge were actively pursuing). Peebles had already shown that throughout the evolution of an expanding, hot, homogeneous, isotropic, model universe up to the decoupling era, the thermal equilibration time was much shorter than the expansion time, therefore necessitating a thermal equilibrium post-decoupling radiation remnant as well. Moreover, that first pair of New Jersey brightness measurements was quantitatively consistent with thermal equilibrium radiation specifically at a temperature of 3K. So if truly blackbody in nature, the peak of the associated spectrum would lie near wavelength A = 1 mm, and consequently the distinct shape of the Planck spectrum might even be observable in the shorter wavelength measurements planned for White Mountain.

The introductory paragraphs of the paper in Physical Review Letters reporting results of the White Mountain expedition (Stokes, Partridge and Wilkinson 1967) emphasize the importance of this test: "To be convinced that one is seeing true blackbody radiation, and not that of a hot gray body, it is necessary to go to shorter wavelengths and look for the curvature in the spectrum due to quantum effects." For 3-K radiation the Planck function lies below the extrapolated long-wavelength power law by 15% at A = 1.58 cm, and nearly 30% at 8.56 mm. Was detection of such departures within the reach of these planned ground-based measurements?

To appreciate quantitatively the challenge posed by this test, and to set out terminology for the following discussion, a brief technical digression seems appropriate, but the reader may skip the next paragraph to find a brief summary in the one after without losing the thread of this tale.

Although the spectrum of the radiation brightness for the blackbody and graybody options under consideration here are properly represented in Figure 4.36, these quantities are not directly observable with a Dicke radiometer. By design the output of our radiometer is proportional to the microwave power coupled into the radiometer by its antenna, and in a preselected, narrow, frequency band Av (centered on v = c/A). For isotropic, thermal-equilibrium radiation, this narrowband power is proportional to the product A2BvAv, where Bv is monochromatic brightness (the Planck function) and the factor A2 enters because the effective solid angle of the antenna pattern is determined by diffraction. Moreover, defining x = hc/Akr, where t is the thermodynamic temperature of the blackbody cavity, the radiometer output is seen to be proportional to A2BV/2k = t[x/(ex — 1)], which has the units of temperature. Thus, when properly calibrated with a blackbody of known t, the radiometer output directly indicates a "temperature" that is proportional to the microwave power detected by the radiometer, and this observable is referred to as antenna temperature. In the Rayleigh-Jeans regime (A ^ hc/k, i.e., for small x) antenna temperature is manifestly constant (independent from A) and equal to the thermodynamic temperature, t. As A decreases, the antenna temperature drops below t just as the Planck function drops below the Rayleigh-Jeans, power-law extrapolation of the Planck function in Figure 4.36; that is, TAntenna = A2BV/2k = t[x/(ex — 1)]. From these definitions, it follows that TAntenna/T ~ 0.7 for t = 3-K black-body radiation at A = 8.56 mm as stated earlier. Consequently, this 30% differential reduction in monochromatic brightness of the blackbody relative to the graybody case is seen in practice to be measured by the radiometer as an absolute reduction of TAntenna below the longer-wavelength value t .

Given that the Dicke radiometer output is calibrated to indicate antenna temperature, TAntenna, and if the CMBR is blackbody radiation with t ~ 3K, testing the graybody null hypothesis against the blackbody alternative, amounts to determining if there is a statistically significant reduction in the antenna temperature measured at A = 8.56 mm compared to the mean antenna temperature (TAntenna ^ t ) associated with the longer wavelength (Rayleigh-Jeans) measurements at A = 3.2 cm and 7.35 cm.

A discriminating hypothesis test would not be possible at A = 8.56 mm with the relatively large water-vapor fluctuations (hence antenna temperature fluctuations) that plague sea-level New Jersey observations. Would rising to an altitude of more than two miles, thereby reducing the vertical column density of water vapor and the rms fluctuation thereof, more than compensate for the increased, intrinsic strength of water vapor emission resulting from observing at such a short wavelength? A definitive answer to this question followed from analysis of the White Mountain data.

Atmospheric emission noise was not the only concern. The 1965 CMBR observations from the roof of Guyot Hall on the Princeton campus (Roll and Wilkinson 1966, 1967) indicated that the error budget for antenna temperature measurements made under the expected improvement in atmospheric conditions at White Mountain would probably be dominated by the uncertainty in instrumental systematic effects, not random atmospheric fluctuations. The mission was doomed unless radiometer design and calibration techniques were significantly improved to reduce the magnitude of these biases, and more precise empirical methods devised to establish quantitatively the associated uncertainties therein.

The White Mountain data subsequently validated both this concern and the effectiveness of the remedies. At wavelengths A = 3.2 cm and 1.58 cm the observed statistical fluctuations from water vapor emission were significantly diminished at higher altitude as expected. Combined with the hard-won reduction in the measurement uncertainty of instrumental sys-tematics, the resulting overall improvement in measurement precision, even at A = 8.56 mm, was gratifying. The time and ingenuity devoted to reducing systematic error associated with these new radiometers had been crucial to achieving the comparatively precise antenna temperatures reported at all three wavelengths, and the overall estimate of the CMBR temperature was tightened to roughly tcmbr = 2.7 ± 0.1 K. Moreover, these successes reduced measurement uncertainty to the point that a significant reduction in antenna temperature at A = 8.56 mm was observed relative to the average of the new, more precise values now established at A = 3.2 cm and 1.58 cm.

A standard test of the graybody null hypothesis can be phrased roughly as this question: given the observed values and their uncertainties, what is the probability that the value of the antenna temperature measured at A = 8.56 mm is not statistically distinguished from the measurements made at substantially longer wavelengths? If that probability is small enough, the graybody hypothesis should be rejected. More specifically, we want to conduct a one-sided test, asking for the probability that the short-wavelength value does not exceed the long-wavelength values in order to reflect the asymmetric aspect of our alternate hypothesis. In this latter formulation, calculating from the radiometric data, the probability that the graybody null hypothesis might be true was about 1/160 = 0.0062, which can be expressed as a confidence of 99.4% (1 - 0.0062 = 0.9938).19 By most standards this is shy of grounds for definitive rejection, but stronger than expected given the apparent measurement difficulties that had been faced. This fortunate outcome was noted in a summary remark in the paper in Physical Review Letters reporting the White Mountain results (Stokes, Partridge and Wilkinson 1967), "We believe this is good evidence for spectrum curvature, and argues strongly against a hot gray-body source."

There were, however, three subsequent indications that a rejection of the graybody hypothesis with confidence of 99.4% was not considered strong enough. First was the decision to proceed with the tentative plan to conduct another high-altitude measurement at an even shorter wavelength, A = 3.3 mm, where the antenna temperature of a 2.7K blackbody is only 1.1 K, or 1.6 K less than a putative graybody source consistent with the longer wavelength measurements. The second is found in the introduction to our paper in Physical Review Letters reporting the A = 3.3 mm result (Boynton, Stokes and Wilkinson 1968): "Recent microwave radiometer measurements at wavelengths of 3.2 cm, 1.58cm, and 8.56 mm have indicated a blackbody of temperature 2.7 °K, but we have not yet demonstrated conclusively the spectral curvature expected from quantum effects." Dave Wilkinson's presentation to the 1979 Copenhagen Symposium, The Universe at Large Redshifts, holds the third (Wilkinson 1980). Regarding the 1967 and 1968 series of ground-based measurements, he writes: "So, the main goal of these experiments was achieved; a gray-body spectrum was ruled out with 5-a significance." In this publication he cites two papers in support of this conclusion, both reporting results at A = 3.3 mm. One is the 1968 observation by the Gravity Group (Boynton, Stokes and Wilkinson 1968). The other, a more precise, corroborating experiment by a group from the Aerospace Corporation (Millea et al.20 1971). The Princeton measurement alone rejected the graybody hypothesis with 99.998% confidence.21 Comparison of a 2.7-K thermal spectrum and its extrapolated Rayleigh-Jeans power law is shown in Figure 4.36, along with the results of the Climax and White Mountain brightness measurements and CMBR

19 This corresponds to rejection at a significance level of 2.5a — the equivalent-single-tail-Gaussian-critical point for a t-test with t = 2.67 on 26 degrees of freedom.

20 The same people had generously provided us with the E-Band diode mounts in the fall of 1967.

21 This corresponds to rejection at a significance level of 4.2a — t-test with t = 5.11 on 25 degrees of freedom.

thermodynamic temperatures. (Partridge presents another discussion of this set of measurements; see Figure 4.28.)

So, if a much stronger test of the "fireball hypothesis" were possible at this shorter wavelength, why wasn't a 3.3-mm radiometer included in preparations for the 1967 White Mountain project?

There were at least two reasons. First, a credible atmospheric emission model with some empirical foundation was used to estimate the magnitude of this challenge. The model indicated that for the same atmospheric conditions (same observing altitude, temperature, water vapor column density, water vapor inhomogeneity, and wind profile) not only would the water vapor emission be five times stronger at A = 3.3 mm than at 8.56 mm, but the absolute antenna temperature fluctuations would also be five times larger (for the same antenna beamwidth). This strong dependence of emission on wavelength is primarily due to the broad tail of the very strong H2O atmospheric line at A ~ 0.6 mm. Even though the antenna temperature difference between the graybody and blackbody hypotheses is 2.3 times larger at A = 3.3 mm, this contrast would be swamped by the still larger increase in water vapor emission fluctuations. Second, at that time there were no reliable, commercially available mixer diodes operating at frequency 90 GHz (A = 3.3 mm), a critical component of the Dicke superheterodyne radiometer. Moreover, completing a mixer-diode research and development program before the summer of 1967 was out of the question, and the expected increase in atmospheric emission noise would seem to make this effort a fool's errand in any case.

As it turned out, after analyzing the White Mountain A = 8.56 mm data, and thereby better understanding the magnitude of water-vapor fluctuations, prospects for a A = 3 mm attempt appeared somewhat more grim. But David Wilkinson had a trick in mind. He determinedly scheduled a second expedition - this time to the National Center for Atmospheric Research (NCAR) High Altitude Observatory at Climax, Colorado in the midst of the Colorado Rockies, a mile higher than Denver ... and in the dead of winter. In fact, he chose the dates according to the historical record to maximize the chance that we would encounter ground-level temperatures of —20 °C or below. His idea was to exploit the precipitation of atmospheric water vapor as ice crystals, precipitously reducing the partial pressure of molecular H2O, and, therefore, the fluctuating emission background as well.

On the basis of this inspired strategy, Wilkinson, Stokes, and I made plans to develop our own mixer diodes, construct a 90-GHz Dicke radiometer, and assemble a reflector and ground shield to be deployed at Climax in March of 1968.

My part in this work brought me to frequent interaction with Bob Dicke -an experience by which I came to appreciate the true greatness of this man. While Wilkinson, Partridge and Stokes toiled on White Mountain, my task was to get up to speed regarding microwave design principles generally, Dicke radiometers in particular, and the handicaps posed at that time specifically by millimeter-wave mixer diodes and klystrons. For the basics I studied Dicke's ground-breaking radiometer-design publication from the 1940s (Dicke 1946b), and several of the MIT Radiation Lab volumes to which he had contributed substantially. Also, I frequently sat with the master himself, whom I found quite helpful and accessible - he relished talking about the "old days" at the Rad Lab. For me, these sessions could be daunting as well as enlightening. Not because of his manner, which was always relaxed and unassuming, but that the depth and range of his knowledge, experience, and analytical powers were sometimes overwhelming. In time, however, I came to admire Bob Dicke for reasons beyond his towering intellect. The following glimpses illustrate for me this man's grace and true virtue.

A few weeks before I joined the Gravity Group that first summer as a fresh-faced postdoc, I answered my home phone one afternoon and heard the greeting, "Hello, this is Mr. Dicke." At first, I didn't recognize who this might be. The only Mr. Dicke I knew was Arnold Dicke, one of my classmates, and this voice was nothing like Arnold's, nor would Arnold have identified himself to me as "Mr. Dicke." Milliseconds later I realized just who was at the other end of the line - a person whom I would have referred to with great deference as Professor Dicke. He congratulated me on my National Science Foundation fellowship award, said that he looked forward to talking with me about choosing a project that would fit into one of the group's current research activities, and asked if I would please stop by his office the next day. Since that incident I have noted for special respect those people who intentionally do not use their titles.

Months later, while walking with Dicke late one afternoon along the basement hallway of Palmer Lab toward the Gravity Group warren, we encountered a longtime building custodian who walked with a pronounced limp. He had always struck me as a pretty rough character with an arresting similarity to my image of Stevenson's Long John Silver. I was somewhat surprised when my professorial companion interrupted our conversation to stop and shake the man's hand. Surprise turned to shock when Dicke greeted the custodian by referring to him as "Gimpy." Bob Dicke and Gimpy (obviously pleased and grinning) exchanged a few humorous jibes before moving on. They seemed an odd duo, but the easily displayed affection and regard for each other engraved on me another life lesson.

From the sidelines I observed many similar encounters. In my mind, Bob Dicke was a giant - deserving all the accolades and personal regard a world can bestow. Yet, in his actions and general disposition, he revealed his own view of himself: a man like any other, obligated to defer to greatness in mind or deed, wherever he found it. I am reminded of Jane Austen's line describing a man of considerable accomplishment and wherewithal who might have grounds for self-important vanity, but whose eventual internalization of egalitarian principles made him an endearing, even though towering, character -"Indeed, he has no improper pride."

As a neophyte, I had the good fortune to know several such men whom I came to view as shining archetypes. Each easily fulfilled that key requirement of greatness: never seeming to live inside it, preferring to occupy the adjacent, less populated ground of humility. In this context I am compelled to recognize another stimulating contributor to Gravity Group meetings, another giant and also someone I greatly admire, John Wheeler.

At that time John's office could be found in that rarified atmosphere of Palmer Lab's upper floor mentioned earlier, and to which I only occasionally ascended. However, while on leave for an academic quarter at the University of Washington in the 1970s, John occupied the office next to mine. Our nearly daily conversations let me know him more closely as an admirable colleague and valued friend. Curiously, just as with Dicke, this vignette also involves a custodian.

John's habit was to leave his office door open while working at his desk. He would close the door, though, when in conversation with a guest or colleague. While meeting with him on many occasions, there inevitably would be a knock at the door. Rather than shout out, "come in," he would jump up from his chair, stepping briskly to the door, always with a warm greeting for whomever awaited him. Like an unusually attentive ER doctor conducting triage, he would quickly answer a question or graciously request they return later. Whether custodian or colleague, John's spirit remained unchanged. The custodian, however, was invited to enter at once in order not to interfere with his duties.

Another exception was his lovely wife, Janette, whom he would invariably ask to be seated - a gentle signal for the current guest to leave, but still with a clear invitation to continue the conversation later. Whether Janette, who unfailingly received his immediate attention, or any of the rest of us, who would be welcomed at a later time, all knew they would receive John's undivided attention . . . until there might be a knock at the door. It was a mystery to me how he could be so selflessly accommodating. This was a man who carried in his shirt pocket five appointment books bound together with a rubber band - one book for each day of the week - so full was his calendar.

By the time Wilkinson, Partridge, and Stokes had returned from White Mountain, Dicke had introduced me to many subtleties of waveguide components. Particularly fascinating was the exploitation of various symmetries in the design of four-port devices, especially the "Magic Tee," and the attendant advantages of a balanced mixer - even though not solving the problem posed by having no known source of reliable mixer diodes.

The fabrication of microwave devices had been gradually improving since the push to develop 3-cm radar during World War II, but in the mid-1960s, 3-mm components were still near the development frontier. I heard from Bill Ernst at the Princeton Plasma Physics Laboratory, who used millimeter-wave instrumentation for plasma diagnostics, and from the microwave research group at the nearby RCA Laboratories, that mixer diodes were not the only weak link in available millimeter-wave hardware.

At that time klystrons were the only practical millimeter-wave local oscillator for superheterodyne receivers and were notoriously inefficient, ran hot, and were noisy due to amplitude fluctuations and mode shifting. This oscillator noise extended out to hundreds of MHz from the line center, and even with a balanced mixer could degrade receiver performance. Both groups suggested we could improve the klystron performance by immersing it in an oil bath. This turned out to be good advice, and we took the additional precaution of choosing the receiver IF band to extend from 0.5 GHz to 1.0 GHz, an order-of-magnitude increase in center frequency and bandwidth relative to the White Mountain radiometers.

For a balanced mixer to work well, a pair of diodes with well-matched characteristics are required; but we faced difficulty procuring or fabricating even a single device that behaved remotely like a diode when placed in a section of waveguide and driven at 90 GHz, let alone two similar ones. Dave Wilkinson kept making phone calls and found that engineers associated with the US Army's Communication and Electronics Command at Fort Monmouth, NJ, were working on just this problem and would meet with us. So Dave, Bob Stokes, and I drove off toward the Jersey shore one late-summer morning. We met with a few casually dressed technical people accompanied (or perhaps monitored) by a couple of officers in uniform. The conversation soon moved to details, and our questions became more penetrating. Because of security concerns the officers suggested it would be more productive for us to talk with certain people at the Aerospace Corporation who could sell us two GaAs point-contact mixer diodes in E-band (90-GHz) waveguide mounts. Although abruptly ending the conversation, this was a good tip.

These mounts (with tunable shorts) turned out to provide a good RF match between waveguide and diode. The resulting receiver performance seemed adequate for our application, but the diodes were mechanically fragile and extremely susceptible to static-discharge damage despite elaborate precautions. Ours survived as a pair less than a week in a relatively benign laboratory environment. Their delicate constitution did not bode well for out-of-doors service amid the rigors of mid-winter weather in the Colorado Rockies.

Desperation was mounting when our friend Bob Wilson called one day to suggest we contact his Bell Labs colleague at Crawford Hill, Dr Charles Burrus, who had been developing millimeter-wave mixer diodes in connection with in-house communication research. Charlie was a godsend - friendly, helpful, and seemingly unfettered by proprietary security concerns. Although our worst fears about having to learn to make our own diodes were confirmed, he explained the entire process in practical terms and sent us back to Princeton with an adequate supply of special materials to remake enough point-contact, Schottky-barrier diodes in our Aerospace Corporation mounts to replace years of failed junctions.

These materials consisted of a few feet of gold-copper wire, from which electrolytically to form sharpened "cat whiskers," and a few dozen tiny, 1mm square, GaAs tiles to be indium-soldered to a post in the diode mount, then chemically etched immediately before attempting to make contact with a carefully inserted whisker.

I include this detail to provide the context for our concern that each time a diode failed at Climax, we would have to carry out this delicate process on the equivalent of a park bench. I say, park bench because the prescribed GaAs etchant was concentrated bromine - tricky to use without a fume hood. We knew our Climax accommodations would certainly not include access to laboratory facilities. Our plan was to conduct diode refurbishing in the great Colorado outdoors, with personnel properly upwind of any alpine breeze.

After much practice remaking diode junctions in Palmer Lab and testing the performance of each as a mixer, we came to the more sobering realization that recovering from a failure probably meant fabricating several new junctions to achieve a reasonable match to the still-functioning mate. We imagined setting up our microscope, chemicals, and etching equipment on top of a packing crate and manipulating tiny parts with gloveless fingers in subzero weather for hours at a time - a sufficiently grim prospect to give even eager postdocs a second thought.

About two months before we were to leave for Colorado, Charlie Burrus, already held by us in high regard, elevated himself to near god-like status by announcing in a phone call that he had, through micro-lithography and a host of other tricks, been able to deposit a fairly dense, uniform array of 2 ^m gold dots on the surface of a pristine GaAs wafer before dicing it into tiles.

With this giant-step innovation, making a diode was now simply a matter of sharpening a whisker and inserting it into the diode mount to make contact with one or another of these gold dots. It might take several stabs to do so, but since Charlie had covered the GaAs surface between the dots with an insulating layer of SiO2, achieving success was unmistakable. Moreover, diode junction performance was found quite uniform from dot to dot. Charlie's remarkable gift meant we would neither need to carry a bottle of bromine to Climax, nor freeze our fingers off. Maybe we could make all this work after all.

Our radiometer noise temperature was typically between 104 K and 2 x 104 K, which may seem high, but implies a one-sigma antenna temperature measurement uncertainty of only 5mK (0.005 K) for a 10-minutes integration time, the smallest single contribution to the error budget for our CMBR observations.

Not surprisingly, the more practiced we became with diode fabrication and handling, the more robust they seemed to be. Although we went well prepared to rebuild diodes at Climax, the initial pair saw us through the entire set of observations.

Perhaps the most important lesson learned on White Mountain was that minimizing calibration errors and the uncertainty associated with various systematic effects could make the difference between success and failure, even if atmospheric fluctuations dominated the error budget. I'm sure this was why Dave made instrumenting the Climax cold load and refining the evaluation of calibration errors his preoccupation during that previous autumn, repeatedly testing and modifying both the hardware and the technique. In addition to work with mixer diodes, receiver fabrication, and testing, Bob and I had measured the side- and back-lobe response of the large conical horn antenna we fabricated by electro-forming copper on an aluminum mandrel, then passivated by gold plating to maintain a high-conductivity (low-emissivity) surface.

To prepare for Climax, we began packing our equipment in early March for airfreight shipment to Denver. The grant bought the whole team war-surplus parkas, but we purchased our own long underwear and suitable cold-weather boots. During the flight west out of Newark, there were moments when my excitement over the possibility of participating in a major scientific advance nearly fell under the wheels of my fear of falling short of that challenge.

As Dave made final arrangements for our travel, I suggested we stay the night in Denver at the Cosmopolitan because a close family friend, Robert Wilhelm, was the hotel manager. I was certain we would find ourselves well hosted, and during the flight I may have raised expectations a bit high. Carelessly, I had not alerted Mr Wilhelm's office until the day before our departure; and by not speaking with him directly was unaware that he had been, and still was, out of town as I strode into the hotel lobby that evening.

Upon checking in I learned of some difficulty regarding the reservation, and we found ourselves escorted to a rather small, disagreeable room looking out into an airshaft. The hotel provided two rollaway beds for Bob and me, leaving little maneuvering space in the already-cramped room. Dave would not let this episode rest, and for the next week or more ribbed me about our wonderful Denver accommodations as a result of my "special" connections.

Not highly motivated to remain in such tight quarters, Dave, a jazz lover, wanted to see what the Denver clubs had to offer. With Bob and me in tow, Dave lurched in and out of Denver's nightspots. In short order, it became clear that Denver was far from New York City in more ways than geographical. The best approximation to a jazz club he could find was actually a strip joint where the dancers lacked attire in the tradition of Kansas City - they'd gone about as fur as they could go. Bob and I flanked Dave at the bar, which also served as the stage, so drinks had to be passed between the dancers' feet. I can speak only for myself, but I'm sure my wide eyes betrayed me -revealing a pitiful attempt to feign ennui - while Dave sat snickering to his Svengali self upon realizing his corrupting influence on at least one of the less worldly fellows at his side. I'm also certain that Bob Dicke never heard of this delightfully tawdry incident.

The next day at the airport we loaded a rented truck with the crates containing our equipment and headed west and upward through deepening snow to the summit of Fremont pass. There, we found Climax, Colorado, at that time the site of NCAR's High Altitude Observatory, the world's largest molybdenum mine, and what had been (prior to 1965) the highest altitude, populated settlement in the USA.

We lived and worked in a simple cabin, set on a hillside with a spectacular view of Sheep Mountain from across a wide alpine valley. Giant columns of ice reached the ground from the roof on either side of the small porch, clearly marking the entrance to our quarters, and a thick snow pack blanketed the entire scene. Most of that first day was spent vigorously shoveling snow from the large concrete pad directly in front of the cabin, where we would set up the aluminum ground shield that would surround our precious radiometer. This wintry setting is shown in Figure 4.37 with the shield and instrument in

Fig. 4.37. The 3-mm radiometer operating near NCAR's High Altitude Observatory at Climax, Colorado. Note the huge icicles on our cabin in the background - this visit was in mid-winter to take advantage of a "freeze-dried" atmosphere. This photo shows the crucial step of calibrating the instrument by measuring the radiation temperature of a microwave absorber immersed in liquid helium contained in the wide-mouth dewar at the lower left protruding through the hole in the ground screen.

Fig. 4.37. The 3-mm radiometer operating near NCAR's High Altitude Observatory at Climax, Colorado. Note the huge icicles on our cabin in the background - this visit was in mid-winter to take advantage of a "freeze-dried" atmosphere. This photo shows the crucial step of calibrating the instrument by measuring the radiation temperature of a microwave absorber immersed in liquid helium contained in the wide-mouth dewar at the lower left protruding through the hole in the ground screen.

the foreground undergoing low-temperature thermal calibration by measuring the radiation temperature of a near-ideal microwave absorber immersed in liquid helium (Boynton, Stokes and Wilkinson 1968). There was also a high-temperature calibration point set by observing an ambient-temperature blackbody. The section of the ground-radiation shield on the camera side has been removed to provide a view of the liquid helium dewar coupled to the horn antenna of the radiometer.

The intense headaches that began for each of us that evening may have been unavoidable at such an altitude, but the day's exertion by normally sedentary workers had not helped. Even though the better part of a week was required to acclimate ourselves, during that time we managed to get things pretty well set up. Mild hypoxia not only slowed us down, but also led to a number of silly mistakes that often triggered spells of uncontrollable giggling - by Bob and me.

Fig. 4.38. This photo, again with the ground-radiation shield temporarily removed from the camera side, not only shows the instrument configuration for determining the atmospheric contribution to the sky temperature, but also renders the various elements of the radiometer more clearly visible: the main horn and the smaller, skyward-pointing, reference horn are seen directly connected to two ports of the microwave switch. A third port opens to an isolator, then on to the mixer diodes of this heterodyne receiver (which are wrapped in foam insulation and directly in front of the klystron oil bath at the far left). The fourth port is terminated with a smaller sky horn as is one port of the isolator. The ferrite switch alternately connects the main and reference horns to the receiver at a 1-kHz rate producing a modulated receiver output whose amplitude is proportional to the difference in microwave power received by these horns. The reference horn is pointed toward a fixed sky region, whereas the main horn beam scans through a range of zenith angles depending on the attitude of the hinged reflector panel to the right.

Fig. 4.38. This photo, again with the ground-radiation shield temporarily removed from the camera side, not only shows the instrument configuration for determining the atmospheric contribution to the sky temperature, but also renders the various elements of the radiometer more clearly visible: the main horn and the smaller, skyward-pointing, reference horn are seen directly connected to two ports of the microwave switch. A third port opens to an isolator, then on to the mixer diodes of this heterodyne receiver (which are wrapped in foam insulation and directly in front of the klystron oil bath at the far left). The fourth port is terminated with a smaller sky horn as is one port of the isolator. The ferrite switch alternately connects the main and reference horns to the receiver at a 1-kHz rate producing a modulated receiver output whose amplitude is proportional to the difference in microwave power received by these horns. The reference horn is pointed toward a fixed sky region, whereas the main horn beam scans through a range of zenith angles depending on the attitude of the hinged reflector panel to the right.

The age gaps among the three of us were not appreciable, but looking back I imagine that Dave saw this regression of our comportment as some cosmic test of his patience. Yet he managed to remain remarkably good-natured throughout our ten days at 11,000 ft.

The closer view of the instrument in Figure 4.38 shows the instrument configuration employed to determine the atmospheric contribution to the sky temperature. The radiometer remained stationary while the hinged panel on the right was placed in a sequence of five attitudes to reflect the horn-antenna beam to view the sky from zenith angles between 0° and 60°. The differential microwave power measurements made in this configuration allow isotropic radiation from above the atmosphere to be distinguished from atmospheric thermal emission (see Roll and Wilkinson 1967). For this purpose the reflector panel replaces the radiation shield with cutout to receive the calibration dewar as shown in Figure 4.37.

The best observing conditions (coldest, driest atmosphere) prevailed during the nighttime. After six days of preparation, we began the observing routine by transferring liquid helium into the cold load in early evening and conducting the interleaved sequence of atmospheric tips and calibrations (as outlined in Wilkinson 1967) until after midnight. This procedure was repeated over the four-day interval near the end of our stay.22

True to best practice, we did not analyze the data before returning to Princeton. Rather, we concentrated simply on following the data-collection routine we had carefully laid out and rehearsed in advance. Always there was the subtle, guiding hand of David who had worked the previous summer at an even higher altitude. From that experience he understood that "best practice" was not only a good idea, but was also forced upon us by the fact that even after nearly two weeks at altitude one cannot think carefully when deprived of oxygen (Figure 4.39). Despite this zombie-like

Fig. 4.39. Hypoxic zombies: David Wilkinson, Robert Stokes, and Paul Boynton.

22 My recollections are admittedly idiosyncratic and myopically focused on the activities of the Princeton Gravity Group reaching to grab the brass ring by striving successfully to test the Fireball Hypothesis. Many other groups were carrying out and publishing CMBR observations prior to the completion of the Climax effort, and their contributions are acknowledged in Figure 2 of Boynton, Stokes and Wilkinson (1968). Of particular relevance to the story I tell in this essay is the elegant exploitation of low-lying rotational states in interstellar CN as a low-temperature cosmic thermometer operating at X = 2.64 mm. This technique has an interesting history reviewed by Thaddeus (1972). The obvious question to ask here is how the papers by Field and Hitchcock (1966) and Thaddeus and Clauser (1966), which were published well before we laid plans for the short-millimeter-wave part of the spectrum, influenced the radiometric study carried out by the Gravity Group in 1968. Taken at face value and treated as upper limits (in view of possible excitation by other processes, which had not been fully evaluated at that time), these fairly tight CN constraints on the monochromatic brightness spectrum clearly favored the Fireball Hypothesis and were considered persuasive by some. I remember some concern regarding biases that could arise in the several inference steps undertaken to move from the observed optical transitions to quantification of excitation by millimeter-wave photons (the technique is rather indirect). This reservation, combined with questions about the local (rather than global) nature of the interstellar gas clouds studied — of which there were only two — kept us from averting our vision.

limitation, we were encouraged by the relatively modest atmospheric fluctuations we recorded. The air above us was certainly cold and apparently quite dry.

Most nights the temperature would drop below —20 °C. Occasionally we would point a flashlight vertically upward. Although the beam was swallowed by the blackness of a moonless sky, one could visually observe the fulfillment of David's prophecy of a dehydrating atmosphere: tiny, rod-like, hexagonal-column ice crystals reflecting glints of light from minuscule facets while tumbling toward the ground - a fascinating phenomenon I had not seen before and not since.

After a day packing up our gear we returned to the airport. Arriving around noon, Bob and I delivered the crates to the airfreight platform and returned the truck. Dave smugly departed immediately on a flight to the east, thinking to leave Bob and me to suffer the privations of another night at the Cosmopolitan, Denver's finest.

Robert Wilhelm, a consummate hotelier, had long returned, and knew that Stokes and I had a reservation at his hotel that night. Upon asking to check in at the desk, we were instead escorted to the manager's wood-paneled office where we sat for some time sipping 50-year-old Tawny Port and chatting amiably about our Rocky Mountain adventure. Wilhelm apologized for not having properly lodged us previously when he was away and personally escorted to us to our rooms, accompanied by a cadre of bellmen. I say "our rooms" because we were lodged in a very large and quite elegantly decorated penthouse suite, replete with vases of fresh flowers and complimentary chocolates. Moreover, Robert and Irene Wilhelm, whom I have known and admired since my early childhood, invited us to a memorable haute cuisine dinner that evening.

David never admitted believing any of this. At some point, though, he had to sign off on the travel expenses for that trip. I've always wondered whether he noticed the absence of hotel and restaurant charges for that last night in the mile-high city where Bob and I snoozed in such splendor.

0 0

Post a comment