SP-4218 To See the Unseen


- Chapter Five -

Normal Science



[117] Starting with the initial detections of Venus in 1961, planetary radar astronomy grew rapidly by discovering the rate and direction of Venus's rotation, by refining the value of the astronomical unit, and by rectifying the rotational period of Mercury. Data gathered from radar observations made at Haystack, Arecibo, and Goldstone formed the basis for precise planetary ephemerides at JPL and Lincoln Laboratory. In sum, the results of planetary radar astronomers served the needs of the planetary astronomy community. In addition, radar also served to test Albert Einstein's General Theory of Relativity.

Planetary radar astronomy concerned itself with two different but related sets of problems. One set of problems related to planetary dynamics and ephemerides, for instance, orbits, rotational and spin rates, and the astronomical unit. A second set related to the radar characteristics, or what is called the radar signature, of the planets, such as surface scattering mechanisms, dielectric constants, and radar albedos. The latter problems are epistemological; that is, they deal with how radar astronomers know what they know.

What defines this second set of epistemological problems is the fact that planetary radar astronomy is based on the use of techniques particular to radar. These problems have remained unchanged over time. In contrast, the first set of problems, those dealing with planetary and dynamics ephemerides, have changed over time. The nature of that change has been additive; at each stage of change, new problems are added to the old problems, which remain part of the set of problems radar astronomers seek to solve.

Both the epistemological and scientific sets of problems are interrelated. For example, planetary radar astronomers derive the ability to solve astronomical problems out of the resolution of epistemological questions. The development of range-Doppler mapping, for example, led to the solution of a set of problems entirely different from ephemerides problems, yet the solution of ephemerides problems was sine qua non to the creation of range-Doppler maps. Conversely, the attempts to solve certain scientific questions required reconsideration of the radar techniques themselves.

The philosopher of science Thomas S. Kuhn has attempted to explain the conduct of scientific activity.1 Although Kuhn has used the term "paradigm" differently over time, initially it had a limited meaning. Stated simply, a paradigm, as used by Kuhn, is a core of consensus within a group of practitioners. The essence of the paradigm consensus is a set of problems and their solutions. Planetary radar astronomy quickly achieved and maintained a paradigmatic consensus on which problems to solve.

Moreover, the field often achieved scientific success by solving problems left unsolved or unsatisfactorily solved by optical means. Just as radar astronomy had resolved earlier that meteors were part of the solar system, so the determination of the rotational rates of Venus and Mercury and the refinement of the astronomical unit were astronomical problems inadequately solved by optical methods, but resolved through the analysis of radar data.

[118] For Kuhn, "normal science" was a specific phase of scientific development distinguished by universal consensus within a given scientific community over the problems to be solved and the ways of solving those problems. In other words, normal science was paradigm science. Preceding its evolution into normal science, according to Kuhn, a scientific activity passes through a developmental phase in which the problem-solving consensus that characterizes normal science does not yet exist. In this "preconsensus" or "pre-paradigm" phase, and immediately before a phase of normal science, groups of investigators addressing roughly the same problems but from different, mutually incompatible standpoints compete with each other. As a consensus emerges, members of the competing schools join the group whose achievements are better, as measured by scientific values.

Planetary radar astronomy did not pass through Kuhn's "preconsensus" phase, however. Complementary, not competing, groups marked the emergence of the field. The "bistatic radar" approach of Von Eshleman at Stanford University complemented the efforts of ground-based planetary radar astronomers, and that complementarity had been Eshleman's intention.2 Ground-based planetary radar astronomers distinguished themselves from the Stanford approach. In a review article on planetary radar astronomy published in 1973, Tor Hagfors and Donald B. Campbell, both at the Arecibo Observatory, explained, "We have, however, chosen to omit this work [space-based radar] here since it is our opinion that it properly belongs to the realm of space exploration rather than to astronomy."3 Space exploration versus astronomy, then, was how planetary radar astronomers established turf lines.

Planetary radar astronomy was, above all else, a set of techniques used with large-scale ground-based radar systems. As a result, planetary radar was an algorithm in search of a problem, a data set in search of a question. Hence, the success of planetary radar inexorably depended on its ability to link its techniques and results to the problem-solving of a scientific discipline. Initially, those problems came from planetary astronomy, but as the types of techniques accumulated, radar came to solve new problems posed by planetary geology. Furthermore, the solving of those problems tied planetary radar astronomy to NASA's space missions.

Despite its mercurial nature, planetary radar astronomy did exhibit an essential characteristic of Kuhn's normal science, a paradigm. The paradigm consisted of a consensus on a particular set of problems (e.g., orbital parameters) and agreement on a particular way of solving those problems (the analysis of range, Doppler, and other radar data obtained with ground-based radars from solar system objects). The detections of Venus, Mercury, and Mars between 1961 and 1963 opened the field, but rotational rates, as well as the refinement of the astronomical unit, established the field. With the successful application of range-Doppler mapping to Venus, the paradigm began to shift in a new direction.


Around the Sun in 88 Days


The first radar detection of Mercury was announced by the Soviet scientists working under Vladimir A. Kotelnikov and associated with the Institute of Radio Engineering and Electronics (IREE) of the Soviet Academy of Sciences and the Long-Distance Space Communication Center near Yevpatoriya, in the Crimea. Kotelnikov's group made 53 radar observations of Mercury during the inferior conjunction with that planet in June 1962. At that time, the distance from Earth to Mercury was between 83 and 88 million [119] kilometers, twice the distance to Venus during inferior conjunction. Although the weakness of the return echoes prevented their use as a reliable indicator of the astronomical unit, Kotelnikov and his colleagues claimed a technical tour de force and a first in planetary radar astronomy.4

Richard Goldstein and Roland Carpenter at JPL took up the Soviet challenge and bounced radar waves off Mercury the following year in May 1963 using the Goldstone experimental radar. The experiment established a distance record that overshadowed the Soviet claim. Mercury was then farther from Earth, over 97 million kilometers away. In addition, the JPL experiment confirmed what astronomers already knew about Mercury, that its period of rotation was 88 days. Goldstein had no reason to believe it was otherwise.5

However, when Gordon Pettengill and Rolf Dyce observed Mercury in April 1965 with the new Arecibo telescope, they reported a rotational rate of 59 ± 5 days. This discovery, one of the earliest major achievements of planetary radar astronomy, astounded astronomers, who sought to explain the new, correct rotational rate. As Pettengill and Dyce concluded, "The finding of a value for the rotational period of Mercury which differs from the orbital period is unexpected and has interesting theoretical implications. It indicates either that the planet has not been in its present orbit for the full period of geological time or that the tidal forces acting to slow the initial rotation have not been correctly treated previously."6

Pettengill, Dyce, and Irwin Shapiro next published a lengthier discussion of their radar determination of Mercury's 59-day rotational period based on additional observations made in August 1965.7 Working with Giuseppe "Bepi" Colombo, an astronomer from the University of Padova visiting the Smithsonian Astrophysical Observatory, Shapiro began to develop an explanation for the new rotational period. Colombo, Shapiro recalled, "realized almost immediately that 58.65 days was exactly two-thirds of 88 days. Mercury probably was locked into a spin such that it went around on its axis one-and-a-half times for every once around the planet. The same face did not always face the Sun. That meant that near Mercury's perihelion, that is, when its orbit is closest to the Sun, Mercury tends to follow the Sun around in its orbit. Near perihelion, then, the orbital motion and spin rotation of Mercury were very closely balanced, so that Mercury almost presented the same face to the Sun during this period."8

In a joint paper, Colombo and Shapiro analyzed Mercury radar data, as well as optical observations from the past, and presented a preliminary model.9 In a seminal paper, Peter Goldreich and Stanton J. Peale pointed out the need to consider the capture of Mercury into the resonant rotation as a probabilistic event. If initial conditions during the [120] formation of the solar system had been slightly different, the capture may not have taken place.10

Irwin Shapiro's graduate student, Charles C. Counselman III, then did his doctoral thesis on the rotation of Mercury. Counselman developed a theory of capture, escape, recapture, and escape, as the eccentricity of Mercury's orbit changed, in a two-dimensional statistical model of the capture problem. Later, Norman Brenner, a graduate student working with both Shapiro and Counselman, expanded the analysis into a three-dimensional model in his 1975 doctoral dissertation. Meanwhile, Stanton Peale published his own three-dimensional analysis.11


The Outer Limits


Although Venus became the prime target of planetary radar astronomers, other planets drew their attention from the earliest opportunity to detect echoes from that planet. Richard Goldstein made the first radar detection of Mars during the opposition of February 1963, when the distance to Mars from Earth was over 100 million kilometers. Goldstein found Mars "a very difficult radar target because of its great distance from Earth and rapid rate of rotation."12

Mars defined the farthest limits of planetary radar detections until after the addition of the S-band radar to the Arecibo telescope and the X-band upgrade of the Goldstone Mars Station. Farther out, neither American nor Soviet efforts ever resulted in an unambiguous radar detection of Jupiter. Certainly no echoes returned from any solid surface features. Nonetheless, US and Soviet investigators claimed detections. The case of Jupiter demonstrates the difficulty of obtaining radar echoes from a "soft" target, that is, one that is not a solid body, especially at such an extreme distance.

Soviet investigators working with Vladimir Kotelnikov at the Yevpatoriya radar center claimed to have detected radar echoes from Jupiter as early as September 1963 in the 29 December 1963 issue of Pravda. The planet was in opposition at a distance of about 600 million kilometers, six times farther than Mars at opposition in 1963. Not surprisingly, Kotelnikov and his colleagues reported that the echoes were weak.13

Between 17 October and 23 November 1963, during the same opposition of Jupiter, Dick Goldstein attempted observations of the planet with the Goldstone experimental radar. He found few if any echoes. Occasionally, though, a single run did indicate a "statistically significant" return. Goldstein noticed that the time interval between these "significant" returns were most often a multiple of the rotation period of Jupiter, about 10 hours. It seemed that a single localized area on Jupiter, which did not coincide with the celebrated red spot, was both a good and a smooth reflector of radar waves.

[121] To investigate further, Goldstein divided Jupiter into eight "time zones" and averaged all the runs which illuminated a single "time zone." The zone centered about the Jovian longitude 32° gave a response that Goldstein characterized as "statistically significant," although, he admitted, "this detection cannot be considered absolutely conclusive." The amount of return was simply too high to be believable. Goldstein later attempted to obtain echoes from Jupiter, using a Goldstone radar that was "a hundred times better," but he did not find any echoes. "We never were able to repeat it," he confessed.14

During the next oppositions of Jupiter, in November 1964, December 1965, and February 1966, Gordon Pettengill, Rolf Dyce, and Andy Sánchez, from the University of Puerto Rico at Rio Piedras, bounced radar waves off Jupiter using the 430-MHz Arecibo telescope. They designed their experiments to duplicate both the Soviet and JPL approaches; however, they failed to validate either the Soviet or JPL claims.

The Arecibo investigators obtained results that were many times smaller than those reported by Goldstein. As for the Soviet results, which were close to the noise level, the Arecibo investigators concluded: "The results reported in the U.S.S.R., which exceed the associated system noise by only 1.3 standard deviations of the fluctuations in that noise, should probably not be taken seriously." The Arecibo investigators suggested that the echoing mechanism was located in the upper levels of Jupiter's atmosphere "and that echoes might be returned only in exceptional circumstances." They concluded: "Many more observations of Jupiter spanning a long period of time and carried out at many widely separated frequencies must be made before the behavior of Jupiter as a radar target can begin to be understood."15

Those observations never took place. Jupiter remained a misunderstood and disregarded radar target. The outer reaches of planetary radar astronomy remained confined to the terrestrial planets. Jupiter and Saturn had to await the Arecibo S-band and the Goldstone X-band upgrades. Even then, however, planetary radar astronomers focused on solid targets, Jupiter's Galilean moons and Saturn's rings.


Icarus Dicarus Dock


In contrast to the attempts on Jupiter, the radar detection of Icarus was unambiguous. Icarus is an Earth-crossing asteroid, meaning its orbit around the Sun crosses that of Earth. On occasion, Icarus comes within 6.4 million kilometers of Earth, as it did in June 1968. Nonetheless, Icarus was a difficult radar target, because of its small size. Its radar detectability was extremely small, one thousandth that of Mercury at its closest approach and only 10-12 (one trillionth) that of the Moon.16

Only Haystack and the Goldstone Mars Station succeeded in detecting the asteroid. Although Icarus was within the declination coverage of Arecibo, attempts on 15 and 16 June 1968 yielded ambiguous results. "A successful search would have been more likely," Rolf Dyce reported, "if the full performance of the line feed had been available."17

Investigators at Haystack Observatory leaped over imposing hurdles to make the first radar detection of Icarus. Irwin Shapiro and his Lincoln Laboratory colleagues prepared an ephemeris based on 71 optical observations of Icarus between 1949 and 1967. Radar observation began in earnest at Haystack on the morning of 12 June 1968. Late that [122] evening, the Haystack observers received a new set of optical positions from astronomer Elizabeth Roemer at the University of Arizona. Michael Ash, of Lincoln Laboratory Group 63, immediately integrated the optical data into the radar ephemeris, and by midnight Haystack was observing with the new ephemeris.

Despite these heroic efforts to organize an improved ephemeris, rain, which severely attenuates X-band radar signals, bedeviled the observations. As a result, Haystack did not obtain a reasonably firm indication of an echo from Icarus until the afternoon of 13 June. Another particularly successful run that evening confirmed the presence of an echo, and by the morning of 14 June success was certain. Haystack terminated observations the morning of 15 June. To achieve its results, the Haystack radar had operated non-stop for 20 hours. Analysis of the data suggested that the radius of Icarus was between 0.8 and 1.6 km.

The effort to detect Icarus in spite of the rain and the difficult nature of the asteroid as a radar target inspired Louis P. Rainville, a Lincoln Laboratory technician who participated in the observations, to compose the following poem:18


Anode to Icarus
Icarus Dicarus Dock
We worked around the clock
For three straight days
We aimed our rays
And an echo showed on the plot.
But as always, there's a woe
The rain made a better show
As bleary our eyes
Stared at the skies
We hoped that the clouds would go.
Oh for the roar and yell
And the glory for old double "L"
If on that crucial day
When it came and went away
We'd had one more decibel!
Now as Icarus speeds from our sphere
These words are for all men to hear
T'was a good show men!
Let's try again ----
In another nineteen years!
And so this was to be our lot
We hoped for more than we got
But we beat the worst;
We did it first!
Icarus Dicarus Dock


[123] Although Rainville's verse implied a contest to detect Icarus, no such competition existed; notwithstanding the rain, the spin direction of the Earth would assure Haystack the first look at the asteroid. At JPL, Dick Goldstein also successfully detected Icarus on 14-16 June 1968. Goldstein used a bistatic radar; the Mars Station received signals from a newly-developed 450-kilowatt transmitter installed on a nearby 26-meter (85-ft) dish. Although the Goldstone transmitter had nearly twice the power of Haystack Observatory, it still received only weak echoes.19

Using optical methods, asteroid astronomers Tom Gehrels, Elizabeth Roemer, and others calculated values for the period and the direction of the spin axis of Icarus and found that it appeared to be a rough stony-iron body, nearly spherical, with nonuniform reflectivity over the surface and with a spin period of 2 hours and 16 minutes. Its radius, they calculated, was at least 750 meters, which was close to the low end of the Haystack estimate. Armed with these results, Goldstein then reinterpreted his radar data and concluded that the surface of the asteroid was rocky and varied in roughness.20

The detection of Icarus was an important achievement of planetary radar, the first detection of an asteroid. Icarus also served to bring together radar and optical planetary astronomers in a special symposium on Icarus organized by Gordon Pettengill and chaired by Arvydas Kliore. Held in Austin, Texas, on 10 December 1968, the symposium was part of the pre-inaugural meeting of the Division for Planetary Science (DPS) of the American Astronomical Society (AAS). Appropriately, the symposium papers appeared in the journal of planetary science Icarus.21

The Icarus symposium was a pivotal moment for both planetary radar astronomy specifically and planetary astronomy in general. Previously, no organization dedicated exclusively to planetary astronomy existed. The AAS had approved the formation of the DPS only a few months earlier in August 1968. In 1973, the DPS opened its ranks to planetary scientists others than AAS members, such as chemists, geologists, and geophysicists, and the DPS endorsed Icarus as the primary publication for planetary research. Under the editorial direction of Carl Sagan, a champion of radar astronomy, Icarus began to solicit more articles in planetary astrophysics, as opposed to the earlier focus on celestial mechanics.22

The Icarus symposium typified the normal science paradigm of planetary radar astronomy in the 1960s. Activity centered on detecting a solar system object with a ground-based radar instrument and analyzing range and Doppler data to obtain information on orbital parameters and radii and related questions. Radar astronomers then presented these results to asteroid astronomers, echoing the fruitful joining of radar observers and astronomers that led to the discovery of the origin of meteors.


The Planetary Ephemeris Program


Starting in the 1960s, the raw data for the improvement of planetary ephemerides was provided by the accumulation of radar range and other data. Traditional observations of planetary positions involved only angular determinations, which provide a position in a two-dimensional plane (the sky). Radar added new dimensions with range and Doppler shift data and included the astronomical unit and the radii and masses of Mercury, Mars, [124] and Venus. JPL and Lincoln Laboratory undertook separate radar ephemerides programs.

The Lincoln Laboratory radar ephemerides program, known as the Planetary Ephemeris Program or PEP, had its roots in the anti-ICBM early warning systems. As a member of a Lincoln Laboratory task force charged with the early detection of incoming enemy ICBMs with radar, Irwin Shapiro became expert in the mathematics of deducing ballistic missile trajectories from radar observations. He wrote up his results in a Lincoln Laboratory report in early 1957. After the launch of Sputnik, the New York publishing house McGraw-Hill released Shapiro's report as a book in April 1958, because his ballistic missile techniques were applicable (with some modification) to satellite tracking. That book then became the basis for the JPL ephemeris program.23

Shapiro and the radar group at Arecibo worked very closely on gathering data for the Lincoln Laboratory planetary radar ephemerides. As Don Campbell explained, "He has always been our ephemerides person, and we provide him with input."24 The close connection between the Arecibo and MIT Lincoln Laboratory groups resulted from the appointment of Gordon Pettengill of Lincoln Laboratory as the first associate director of the Arecibo Ionospheric Observatory. Pettengill set up the program so that the Arecibo radar ephemerides would always come from the PEP group.

Acquiring input for the PEP required extensive data taking that involved long hours of observations, often late at night. Don Campbell and Ray Jurgens, both graduate students at the time, did a lot of the work on Venus, Mars, and Mercury, under the supervision of Rolf Dyce and Gordon Pettengill. Campbell remembered the Mars observations in particular:25

This involved a lot of late nights, unfortunately, because the Mars opposition was around midnight. Every time the radar system was used, you had to go up to the suspended platform and actually change the receiver over. Then you had to go up after you finished to change them over again. Since I was very much at the lowest end of the totem pole at the time, it was my job to get on the cable car, go up to the structure, dabble with the thing late at night, change the receivers, come back, then when we finished, go back up and change them again. I suppose in retrospect you think of it as painful, although at the time I don't remember being particularly worried about it. I probably thought it was fun initially, although there were a lot fewer fences and safety devices on the platform then than there are now. It was quite possible to fall right through the thing.

The initial PEP calculations performed with the planetary radar data served to refine the astronomical unit. Shapiro, however, also saw the need to refine the planetary ephemerides and the planetary masses. "It was also clear to me," he explained, "that we should not do it the way astronomers did, that is, with analytical series expanded out to huge numbers of terms. It seemed to me that with computers, even with those available at that time, we should be able to do this numerically, integrating the equations of the motions of the planets, integrating the partial derivatives, and doing everything digitally."26

The PEP required a large computer as well as an immense computer program. Today, the program has well over 100,000 Fortran statements. Computer programming, [125] however, was not Shapiro's forte. "I am pretty much a computer ignoramus," he confessed. So he hired a summer student, Michael E. Ash, who was a Princeton graduate student in mathematics. After graduating from Princeton, Ash worked at Lincoln Laboratory for about twelve years before taking a position at MIT's Draper Laboratory. Ash was the chief architect of the PEP computer program. John F. Chandler, a graduate student of Shapiro, took over the PEP from Michael and worked on it for over twenty years. Chandler expanded its applications so that now, in the words of Shapiro, "it does everything but slice bread."

Originally, PEP also analyzed optical observations of the Sun, Moon, and planets, including optical data from the U.S. Naval Observatory back to 1850. "I spent more time than I care to admit," Shapiro confessed, "transferring to machine-readable form all the optical observations recorded in history since 1750 of the Sun, Moon, and planets. In the end, I didn't think it was worth it. I never published our results, to Michael Ash's chagrin. We had this manuscript about so high [nearly seven and a half centimeters or three inches], but I could never find enough time to polish it to my satisfaction. History passed us by. That was the biggest unfinished task of my life. Michael Ash put in a lot of work on that, though not as much as I did. But the ball was in my court to finish it off, and I did not do it. So this is a guilt session."27

Today the PEP is a very complicated program that analyzes a variety of observations, including lunar laser ranging data. When he moved to Draper Laboratory, Michael Ash modified it for satellite and lunar work. It is still used in planetary radar and by astronomers at the Harvard-Smithsonian Center for Astrophysics. It can process pulsar as well as Very Long Baseline Interferometry (VLBI) observations. For a while, most of the pulsar observers in the world used the PEP; however, they shifted to the JPL ephemeris program in recent years. A lack of funding has left the PEP just able to keep up with the Arecibo ephemeris work.

In contrast, JPL has had the manpower and funding to support it. JPL developed its radar planetary ephemerides to support NASA spacecraft missions. Today, the JPL planetary ephemeris program, under the direction of E. Myles Standish, Jr., employs about a half dozen people who work on planetary, lunar, cometary, asteroidal, and satellite ephemerides. JPL initially called their ephemeris programs DE followed by the version number, with DE standing for "Development Ephemeris." In the late seventies, JPL sent over 50 copies of its ephemeris DE-96 to observatories, space agencies, and astronomical research groups around the world.

Next came the DE-200 series, which used a new equator and equinox. All major national almanac offices, including the U.S. Naval Observatory, and the French, British, German, Japanese, and Russian almanac offices, now use the JPL DE-200 ephemeris program, as do many universities, the European Space Agency, and radio astronomers. Moreover, the DE-200 program, formerly available on magnetic tape, now is distributed through the Internet as an FTP file.28

The Lincoln Laboratory and JPL planetary ephemeris programs were uses of planetary radar data that did not necessarily lead to publications. Moreover, the vast amount of data routinely collected by radar astronomers and stored in the data bases of those ephemeris programs did not result from experiments designed to achieve a special purpose. Many planetary radar experiments quickly became routine operations. A glance at the extant Haystack radar log books indicates that radar astronomers rarely ran experiments themselves; expert technicians, like Haines Danforth and Lou Rainville, operated [126] the radar equipment, and the software consisted of "cookbook programs."29 This routinization of experimentation is one aspect of Kuhnian "normal" or paradigm science.


Testing Albert Einstein


According to Shapiro and his colleagues at Lincoln Laboratory, the main purpose in gathering radar data for the planetary ephemeris program was "to test Einstein's theory of General Relativity."30 The Shapiro test of the gravitational time delay predicted by General Relativity is interesting for its contribution to theoretical physics and astrophysics, as well as a major early achievement of planetary radar. Its development underscores the close and necessary connection between the capabilities of radar instruments and the kinds of scientific problems that one can solve with radar. It also illustrates the emotional intensity with which scientists struggle to assert their claims of discovery and priority of publication.

After announcing his special theory of relativity in 1905, Albert Einstein spent another ten years developing the theory of General Relativity.31 The theory of General Relativity traditionally has found support in three principal experimental areas. The first came from its accounting for the precession of Mercury's perihelion, the point at which Mercury is closest to the Sun. Traditional theoretical physics had been incapable of explaining the precession of Mercury's perihelion without leaving certain discrepancies unexplained. The ellipse of Mercury's orbit was turning faster than traditional physics said it ought to by an amount of some 43 seconds of arc per century. Einstein found that his equations gave just that amount of deviation from the measure predicted by traditional physics. The perihelion motion came out not only with the right numerical value but also in the correct direction.

Einstein's theory of General Relativity predicted that a gravitational field would bend or deflect the path of light rays. For a light ray glancing the Sun, the theory of General Relativity predicted a deflection of 1.7 seconds of arc, about 1/1,000 of the angular width of the Sun as seen from the Earth. The theory of General Relativity also predicted that the gravitational field would cause the speed of a light wave to slow.

Three types of experimental tests conducted over several decades confirmed the precession of Mercury's perihelion, the deflection of light rays in a strong gravitational field, and the red shift. Consequently, Irwin Shapiro called his the Fourth Test of General Relativity. Initially, Shapiro was interested in using radar to confirm the precession of Mercury's perihelion. He hit upon that idea in 1959, but Shapiro was not sure whether a check on a widely accepted physical theory would be a worthwhile experiment. So in April 1960, he asked a visiting French physicist, Cyrano de Dominicis, about the experiment. [127] Dominicis told Shapiro he thought the experiment worth doing, because scientists had so few tests of relativity.32

No radar at the time, however, had the necessary sensitivity to carry out the precession experiment. At any rate, the Fourth Test was not to measure the precession of Mercury's perihelion, but the slowing down of light waves caused by solar gravity. The new idea came to Shapiro in the spring of 1961, as he was attending a briefing for the military on some of the research conducted at Lincoln Laboratory and MIT with Department of Defense funds. After his lecture on measuring the speed of light, George Stroke, in a conversation with Shapiro, mentioned that the speed of light is not the same everywhere, but depends on the gravitational field through which it is passing. Shapiro was surprised. He refreshed his memory on General Relativity and realized that there was a misunderstanding: according to General Relativity, a (freely) falling observer would measure at any location the same speed of light, independent of the (local) gravitational field. However, Shapiro reasoned, the effect of the gravitational field on the speed of light would be cumulative over a round-trip path (unlike the red shift) and that a radar experiment, therefore, ought to be able to detect this gravitational time delay.

Shapiro now had the idea of testing the gravitational time delay predicted by General Relativity, but he realized that extant radars could not measure this small relativistic effect. Moreover, Shapiro did not write up the idea at that time. "I just kept it in the back of my mind," he explained.33

The inauguration of the Arecibo Ionospheric Observatory in November 1963 revived Shapiro's interest in testing General Relativity. In July 1964, Shapiro and his wife, pregnant with their first child, travelled to Arecibo at Gordon Pettengill's invitation to spend the summer working at the AIO. When Charles Townes, then MIT Provost, visited Arecibo that summer, Shapiro briefed him on his proposed relativity test and told him that Arecibo could not perform the test. "We would never be able to see this effect," Shapiro explained. "The plasma effect of the solar corona would be of the same general type, and the variations would be much larger than the relativistic effect we were looking for. We would never be able to pick it out."34

Shapiro then returned home and learned that Haystack was to be dedicated in October 1964. Suddenly it occurred to Shapiro that Haystack might have enough capability to do the experiment. He did some quick back-of-the-envelope calculations and concluded that Haystack might be able to do the experiment. Shapiro sent his manuscript to Physical Review Letters, and the journal received it on 13 November 1964.35

From his realization that Haystack could do the experiment to his submission of the paper took only one week. After doing the calculations more accurately, Shapiro realized that the sensitivity of the Haystack radar was not good enough to detect the relativistic effect. He and his Group Leader then requested an upgrade of the Haystack radar from the head of Lincoln Laboratory, Bill Radford, who subsequently obtained a funding [128] commitment from the Rome Air Development Center for the upgrade. The upgrade consisted of design and construction of a new electronic plug-in unit, boosting the continuous-wave transmitter from 100 to 500 kilowatts, and replacement of the cooled parametric amplifier with a lower noise maser.36

In January 1965, as the design and construction of the radar upgrade was underway, a colleague showed Shapiro a JPL internal publication dated 31 October 1964 in which an article by Duane Muhleman discussed using radar to measure the general relativistic effect.37 Shapiro was upset. He recalled vividly that in January 1964, he was walking near Harvard Square with Muhleman. When Muhleman asked him why he was still interested in radar astronomy, Shapiro told him about his idea to test this new effect predicted by General Relativity. Yet Muhleman did not acknowledge that conversation in his JPL report. Furthermore, that report only discussed the test being done near the inferior conjunction of Venus, where such a test was, and remains, infeasible. Shapiro noted that several years later, he approached Muhleman's co-author of the JPL report, Paul Reichley, and asked him how he got involved in that project. To Shapiro's amazement, Reichley responded directly that Muhleman had said to him, "Shapiro says there's an effect here, let's look into it."

The Muhleman and Shapiro relativity experiments both involved using radar and finding the relativistic time delay, but the design of their experiments differed widely. The Shapiro Test sent radar waves from Earth to graze past the Sun and bounce from Mercury (or Venus) at superior conjunction, that is, as the planet was just going behind the Sun (or emerging from behind the Sun) when seen from Earth.

The radar waves then returned from Mercury (or Venus) and again passed near the Sun on their return trip to Earth. The Sun's gravitational field would slow down or delay the radar waves. General Relativity predicted that the cumulative time delay due to the direct effect of the Sun's gravitational field might be somewhat more than 200 microseconds. On the other hand, this time delay for radar waves bounced from, say, Venus at its inferior conjunction amounted to only about 10 microseconds.38

Muhleman's experiment grew out of his theoretical work at JPL on communications with spacecraft flying near the Sun. Spacecraft navigation was at that time essentially a matter of measuring Doppler shift to a high degree of accuracy. Because JPL also was considering ranging systems, Muhleman was studying the effects of the solar corona on both Doppler and range signals. "While working on that problem," he explained, "I realized that the main effect of the solar corona on the radio signal was that the signal was bent as it went around the Sun." Muhleman considered the solar gravitational field as though it were a lens with an index of refraction, an idea he later discovered in various relativity books. On a practical level, the Muhleman and Shapiro relativity studies differed widely. Whereas Shapiro intended to bounce radar waves off Mercury (or Venus) at superior conjunction, Muhleman proposed measuring at inferior conjunction, when the relativistic effect would not be detectable.39

[129] The judgement of general texts is that Irwin Shapiro originated the Fourth Test.40 Muhleman, for a number of reasons, dropped out of radar astronomy for over twenty years. Shapiro and his Lincoln Laboratory coworkers eventually did perform the Fourth Test at Haystack during the superior conjunction of Mercury in November 1966. Haystack made subsequent measurements during the superior conjunctions of 18 January, 11 May, and 24 August 1967. The results confirmed General Relativity to an accuracy of about ten percent.41

Additional observations of Mercury and Venus made at both Haystack and Arecibo during several superior conjunctions helped to refine the Fourth Test results. Subsequent experiments carried out on spacecraft further improved the accuracy of the test. The best accuracy yet achieved was from a combined MIT and JPL experiment on the Viking mission to Mars; it confirmed Einstein's theory of General Relativity to a tenth of a percent. The accuracy of the measurement of the relativistic effect had improved by an impressive factor of 100, or two orders of magnitude, in 10 years.42


A Shifting Paradigm


The planetary radar research discussed up to this point shared a consensus on problem-solving activities in a way typical of a Kuhnian paradigmatic science. Among the forces driving the evolution of planetary radar astronomy was the interaction between the two kinds of problems radar astronomers attempted to solve. One set related to the larger theoretical framework which the results of radar observations and analysis attempt to address; the other related to epistemological questions and included radar techniques. Because the two problem sets are necessarily linked to one another, the invention or adaptation of new radar techniques impacted on the kinds of scientific problems addressed by radar astronomy and, as a result, expanded the paradigm without altering the original problem-solving activities and techniques.

One of the most powerful new radar techniques was planetary range-Doppler mapping. It added a whole gamut of answers that radar astronomy previously could not provide. The successful application of the new technique depended on the availability of a generation of highly sensitive radars, Haystack, Arecibo, and DSS-14. Technology continued to drive radar astronomy. Because the kinds of problems range-Doppler mapping solved were related more to geology than to astronomy, planetary radar grew close to the theoretical framework of planetary geology. This shift of the paradigm (without alteration of the original astronomy-oriented paradigm) also reflected the evolving social context of planetary radar, which in 1970 found itself a patron in NASA and its missions of planetary exploration. Thus, changing problem sets and theoretical frameworks on the one hand and the evolution of financial and institutional patronage on the other became inextricably linked.


Planetary Range-Doppler Mapping


Both range and Doppler were standard radar measurements long before they united to provide range-Doppler maps of planetary surfaces. Range or time-delay [130] measurements determine how far away a target is by the amount of time the echo takes to return to the radar receiver. The greater the distance to the target, the longer the echo takes to appear in the receiver. Conversely, the shorter the distance to the target, the less time the echo takes to appear in the receiver. Knowing that radar waves travel at the speed of light, one can calculate the distance traveled by a radar signal from the amount of time between transmission of a radar signal and reception of its echo.

If one assumes that a planetary target is a perfect sphere, then when a transmitter directs radar waves at it, the waves arrive first at a circular area at the center of the planet as viewed from Earth. The point on the planet's surface that has the radar at its zenith, and is thus closest to the observer, is called the subradar point. Thus, the radar waves first hit a circular area on the planet surrounding the subradar point and form what is called a range ring. Within each range ring, the distance from Earth to the planet's surface, that is, the range or delay in time, is the same. The longest delays (and therefore ranges) generally correspond to echoes from near the planetary limbs.

When a radar transmits, it sends a signal that contains only a very narrow band of frequencies and appears almost line-like. Such would be the case, too, for the echo received back at the radar were there no difference in the relative motion between the radar and its target. In reality, when looking from the Earth at a planetary target, this relative motion is always a factor. The combined motions of the Earth as it spins on its axis and orbits around the Sun, and of the planetary target as it also spins on its axis and orbits about the Sun, cause what is known as the Doppler effect or Doppler shift, which is the difference between the frequencies of the radar transmission and the radar echo. The differences in the relative motions of the radar and the target broaden the frequency of the returning signal. Instead of a (nearly) single frequency, the returning signal exhibits a spectrum of frequencies "shifted" or set off from the transmitted frequency.

In order to remove the Doppler shift caused by the relative motion of the observer and the target, planetary radar astronomers generally use a radar ephemeris program. The program automatically adjusts the incoming signal for the expected Doppler shift, which itself changes over time because of the changes in relative motion of the observer and the target. Thus, the predicted Doppler shift must be accurate enough to avoid smearing out the echo in frequency. This requirement places stringent demands on the quality of the observing ephemeris. Thus, the Lincoln Laboratory PEP and the JPL Development Ephemeris series were of vital importance to the successful execution of planetary range-Doppler mapping.

A given portion of the echo frequency spectrum corresponds to a slice or strip on the planet's surface. Each slice is parallel to the plane containing the line from the observer to the planet and the spin axis of the planet, and each slice has the same Doppler shift value, because each portion of that slice of the planet's surface has the same motion relative to the observer. When Doppler shift and range data are combined, the slices of equal Doppler shift intersect the range rings to form "cells." In general, each range-frequency cell corresponds to two particular areas on the planet's surface. The amount of surface area corresponding to a particular range-frequency cell represents the resolution of the radar image on the planet's surface and varies over the planetary surface (see Technical Essay).

The amount of power returned from the target for each range-frequency cell can be converted into a two-dimensional image of the planetary surface through a series of complex mathematical manipulations. Each spectrum has the attributes of power, bandwidth (the maximum spread of line-of-sight velocities), shape, minor features, and a weak broadband component. The total power received depends on such instrument factors as transmitter power, antenna gain, and pointing accuracy, as well as on the reflecting ([131] backscattering) properties of the target and the so-called radar equation. The radar equation states that the amount of power in an echo is inversely proportional to the fourth power of the distance. That means that the echo power received by a radar decreases sharply with increasing distance of the target. For instance, a radar echo is 1/16th as strong if the distance to the target is doubled, all other conditions being equal.

In general, rough planetary surfaces backscatter more power from near the planet's limbs than a corresponding smooth sphere. Thus, a rough surface leads to a broadening of the frequency spectrum of the planetary echo. A smooth planetary surface (relative to the size of the wavelength of the radar waves) would broaden the return signal to a far lesser degree. The amount of power returned to the receiver in each resolution cell therefore corresponds to the planet's surface characteristics.

The idea of combining range and Doppler data to form a radar image came together at Lincoln Laboratory in the late 1950s. There, Paul Green began to consider the calculation of a planet's spin velocity from a simultaneous measure of range and Doppler spread. Another member of the Lincoln Laboratory radar group, Roger Manasse, pointed out that when you look at a spinning object, the planes of equal Doppler shift are parallel to the plane containing the line of sight and the rotation vector.43 However, Manasse did not put the slices of Doppler shift together with the range rings. The originator of that idea was Paul Green.

Green remembers how the idea came to him: "I was sitting in my living room wondering what the relationship was between the two of them. I also had noticed that Ben Yaplee had actually measured those things."44 Ben Yaplee and others at the Naval Research Laboratory were using range rings to refine the Earth-Moon distance. They discovered that details in the structure of the return echoes could be correlated with lunar topography.45

However, they did not develop planetary radar range-Doppler imaging. "It was simply an unevaluated measurement," Green explained. "There was no attempt to know what deeper message might be behind that. I was just thinking, 'Hey! Wait a minute! That's kind of an interesting thing to do.' Maybe it was obvious, but might there not be something deep behind it?"46

Soon after formulating range-Doppler mapping, Green discovered that classified military radar research at the University of Michigan had led to the conception of a similar process but with significant differences. The military process involved imaging the Earth from aircraft and relied on developing a radar "history" of the target to create an image, while planetary range-Doppler mapping created a "snapshot" of a planetary surface from a ground-based radar. Because of the similarities in the two methods, Green was careful to call his planetary range-Doppler mapping. The University of Michigan radar effort, as we shall see in the next chapter, eventually had a profound impact on planetary radar astronomy.

Paul Green first presented his ideas on range-Doppler mapping at the pioneering Endicott House Conference on Radar Astronomy, then at the URSI workshop on radar astronomy that followed immediately afterward in San Diego. An abstracted form of that paper and the others presented at the workshop soon appeared in the Journal of Geophysical Research.47

[132] Green did not apply his theory to actual radar mapping of the planets. Instead, it was his Lincoln Laboratory colleague Gordon Pettengill who used it beginning in 1960. Initially, Pettengill explored the surface of the Moon with the Millstone radar. The result was an image that barely resembled the lunar surface. Pettengill concluded, "It is obvious that much patient work lies ahead before detailed correlation with optical photographs may be attempted."48


Figure 16. The first range-Doppler image of the Moon, 7 January 1960, made by Gordon Pettengill, using techniques developed by his Lincoln Laboratory colleague Paul Green.

Figure 16. The first range-Doppler image of the Moon, 7 January 1960, made by Gordon Pettengill, using techniques developed by his Lincoln Laboratory colleague Paul Green. The top of the image (shown in range box 2) represents the point on the lunar surface closest to the radar. Pettengill, as the first associate director of the Arecibo Ionospheric Observatory, as it was then called, later guided range-Doppler imaging of the Moon and planets at Arecibo as well as at the Haystack Observatory. (Courtesy of MIT Lincoln Laboratory, Lexington, Massachusetts, photo no. 261209-1D.)


[133] Lunar Radar Mapping


Pettengill made a second attempt at lunar radar mapping in June 1961, again using the Millstone radar. Those and the previous Pettengill radar images had what radar astronomers call north-south ambiguity. The nature of range-Doppler mapping is to create an uncertainty (called north-south ambiguity), such that the observer does not know from which hemisphere the echoes are returning. The range-Doppler technique creates two points, one in the northern hemisphere and the other in the southern hemisphere, with exactly the same range and Doppler values. The radar data cannot distinguish the hemisphere of origin of the return echo and thus presents a confusing picture of the target's features.

Pettengill had no technique yet for distinguishing northern-hemisphere echoes from southern-hemisphere echoes. He knew that the youngest, and therefore the roughest, large feature on the lunar surface visible from Earth was the crater Tycho. During a full Moon, this crater appears to have rays emanating across the lunar surface. When Pettengill looked at the echo spectra, he found anomalously high spikes that were consistent from run to run. He assumed that they were in the southern hemisphere (the location of Tycho) and found they matched the crater's location.49

For the first time, a lunar surface feature and a radar return had matched. However, the north-south ambiguity problem stood in the way of refining range-Doppler mapping into a useful tool for exploring the solar system. One solution appeared when the Arecibo Ionospheric Observatory began operation in November 1963. There, at the instigation of Pettengill, who was now Associate Director of Arecibo, Thomas W. Thompson, then a Cornell graduate student, and Rolf Dyce began range-Doppler mapping of the Moon.

In contrast to Millstone, the Arecibo radar antenna had a narrow beamwidth relative to the angular size of the Moon. The Moon has a diameter of a half degree or 30 minutes of arc. The width of the Arecibo antenna beam was 10 minutes. Instead of aiming the antenna at the center of the lunar disk facing Earth, Thompson and Dyce aimed it at a point 10 minutes of arc south from the center. The Arecibo telescope received echoes, therefore, only from the lower or "southern" part of the Moon. The technique assuaged the problem of north-south ambiguity, but was applicable to only the Moon. Venus was only a speck, slightly more than one minute of arc, compared to the Moon's 30 minutes of arc.50

Using this approach, Thompson and Dyce explored eight regions of the lunar surface and collected data on echo strength. They converted the data into "contour" lines of relative reflectivity. Thompson placed these lines, computed and plotted on a transparent overlay, over lunar maps made from photographs. The resultant radar contour map had a resolution of 20 by 30 km.51

Thompson continued to carry out range-Doppler mapping of the Moon by taking advantage of the increasingly narrow beamwidth of the Arecibo antenna. By reducing the beamwidth from 10 to 7 minutes of arc, he succeeded in creating a range-Doppler map of the crater Tycho with surface resolutions between 7 and 10 km. The output from a given radar observation now represented a considerable quantity of data; between 10,000 (104) and 100,000 (105) values of intensity (or pixels) constituted a single map.



Figure 17. Radar map of the lunar crater Tycho with a resolution of 1 kilometer made with the 3.8-cm (7,750-MHz) Haystack Observatory radar.

Figure 17. Radar map of the lunar crater Tycho with a resolution of 1 kilometer made with the 3.8-cm (7,750-MHz) Haystack Observatory radar. The grid lines are spaced about 17 km apart. (Courtesy of MIT Lincoln Laboratory, Lexington, Massachusetts, photo no. 242336-1.)


At the same time, Gordon Pettengill guided lunar radar observations at Haystack, which had become available in late 1964. Haystack, moreover, had a narrower antenna beamwidth, only 3 minutes of arc, and the higher operating frequency of Haystack (3.8 cm, X-band) compared to Arecibo (70 cm, UHF) helped Haystack to achieve a much finer resolution on Tycho: between 1 and 2 km. The Haystack radar images now approached the quality of lunar photographs made from Earth. In the words of Pettengill and [135] Thompson, "The most immediately striking feature of the 3.8-cm [Haystack] observations is their resemblance to the optical photograph...."52

The coincidental refinement of lunar range-Doppler imaging and the commitment to place an American on the Moon before the end of the 1960s enhanced the value of the lunar radar work done at both Arecibo and Haystack. NASA Apollo mission staff used the radar images to help select landing sites, and Apollo funded Thompson's dissertation and subsequent radar studies of the Moon. Once the resolution of radar images surpassed the resolution of lunar photographs made from Earth, the value of lunar radar studies to NASA grew even more. Thus, the new technique brought radar astronomy closer to the scientific needs of NASA, increasingly the patron of radar astronomy.

At Arecibo, Tommy Thompson and Rolf Dyce undertook radar mapping of the Moon at both 40 MHz (7.5 meters) and 430 MHz (70 cm) under a supplementary grant from NASA. A joint report with Lincoln Laboratory compared the Arecibo results with those carried out at Haystack by Stan Zisk with additional NASA funding under a contract between MIT and the Manned Spacecraft Center in Houston. NASA funded lunar studies at both telescopes until 1972, when the Apollo program came to an end.53


Venus Radar Mapping


In 1964, as Thompson and Zisk were starting their lunar mapping activities, Roland Carpenter and Dick Goldstein analyzed spectra from Venus and discovered the first features on that planet's surface. The Goldstone Venus radar lacked sufficient sensitivity to apply range-Doppler mapping to Venus. However, once the Mars Station became available, Goldstein continued his exploration of Cytherean surface features using range-Doppler techniques, but without resolving the north-south ambiguity. Thus began one of the most long-lived and extensive activities of planetary radar astronomers. This scientific niche for radar resulted from that planet's opaque atmosphere which barred exploration with optical methods.

When Dick Goldstein observed Venus during the 1964 inferior conjunction, he looked only at the structure of the spectra returned from the planet. This was the same technique that Roland Carpenter had used earlier to discover the retrograde motion of Venus; it was not range-Doppler mapping. A few topographic features were visible as details in the return spectra. Goldstein found two features represented as peaks. They moved slowly across the spectrogram, a graph plotting echo power density versus frequency, from the high-frequency side to the low-frequency side, in synchronization with the planet's rotation.

Goldstein then placed his two features on a coordinate system with the first feature, named Alpha (Greek letter alpha), located on his zero degree meridian in the southern hemisphere. The second feature, named Beta (Greek letter beta), Goldstein placed in the northern hemisphere. His coordinate system was somewhat arbitrary out of necessity, as astronomers generally had not agreed upon any Cytherean coordinate system. Additional analysis of the 1964 data revealed three more features around the equator. Goldstein named them Gamma (Greek letter gamma), Delta (Greek letter delta), and Epsilon (Greek letter epsilon).



Figure 18. One of the earliest range-Doppler images of Venus made by Richard Goldstein of JPL with the Goldstone radar.

Figure 18. One of the earliest range-Doppler images of Venus made by Richard Goldstein of JPL with the Goldstone radar. The notation "0°" indicates the meridian in Goldstein's coordinate system. Visible are the first surface features identified by Goldstein: Alpha (Greek letter alpha), on the meridian in the southern hemisphere, Beta (Greek letter beta), in the far west of the northern hemisphere, and Delta (Greek letter delta), just to the north of Beta. Gamma (Greek letter gamma) and Epsilon (Greek letter epsilon), two additional features identified by Goldstein, are not labelled. The radar names Alpha and Beta were retained when astronomers began naming the surface features of Venus. (Courtesy of Jet Propulsion Laboratory, photo no. 331-4849AA.)


[137] Although he judged that these features were probably mountain ranges, Goldstein had insufficient evidence. What were they? "Venus is still a mystery planet," Goldstein concluded. "However, it may no longer be viewed as featureless, but rather as an exciting object for further study."54

Using the data taken with the newly operational Mars Station during the 1967 Venus inferior conjunction, Goldstein studied the Beta region in more detail, attempting to determine its size and character, rather than searching for new features. The Mars Station, moreover, provided sufficient sensitivity to attempt range-Doppler mapping. Goldstein observed Beta, Delta, and an unnamed region at (his) 40° South latitude and made a crude radar image of the Greek letter beta region. Still, Goldstein lacked sufficient data to determine whether Beta was a mountain range or another type of feature.55


Figure 19. A detailed radar view of the Beta region of Venus, 1967, made by Dick Goldstein of JPL using the Goldstone radar.

Figure 19. A detailed radar view of the Beta region of Venus, 1967, made by Dick Goldstein of JPL using the Goldstone radar. It exemplifies the limits of resolution available in some of the earliest radar images of that planet. (Courtesy of Jet Propulsion Laboratory, photo nol. P-8882.)


[138] Meanwhile, Roland Carpenter, who was now both a JPL employee and an instructor in the Department of Astronomy of the University of California, Los Angeles, had analyzed 1964 Venus inferior conjunction radar data. Carpenter found two distinct peaks in the return spectra that persisted day after day and moved slowly with time. On closer examination, the first peak appeared to have three components, which he hesitated to interpret because he felt their nature could not be determined with the available data.

Using Goldstein's coordinate system, Carpenter began to identify the most pronounced features with letters of the alphabet from A to G. He labeled less probable locations as numerical extensions of nearby features, e.g., B1, C1, C2, D2, and D3. Correlations between Carpenter's and Goldstein's features began to emerge. Carpenter's feature F had the same location as Goldstein's Greek letter alpha, and Carpenter's group B, C, and D corresponded to Goldstein's Greek letter beta (Table 2).56


Table 2. Radar Features of Venus.








Haystack I








Haystack IV








Haystack II




Haystack III








Haystack A (later Haystack VI)












Haystack B








Haystack V
















Haystack C




Haystack D







R. M. Goldstein, "Radar Studies of Venus," in Audoin Dollfus, ed., Moon and Planets (Amsterdam: North-Holland Publishing Company, 1967), pp. 126-131; R. M. Goldstein and H. C. Rumsey, Jr., "A Radar Snapshot of Venus," Science 169 (1970): 974-977; R. L. Carpenter, "Study of Venus by CW Radar: 1964 Results," The Astronomical Journal 71 (1966): 142-152, especially pp. 148-151; A. E. E. Rogers, T. Hagfors, R. A. Brockelman, R. P. Ingalls, J. I. Levine, G. H. Pettengill, and F. S. Weinstein, A Radar Interferometer Study of Venus at 3.8 cm, Technical Report 444 (Lexington: Lincoln Laboratory, 14 February 1968); A. E. E. Rogers, R. P. Ingalls, and G. H. Pettengill, "Radar Map of Venus at 3.8 cm Wavelength," Icarus 21 (1974): 237-241; D. B. Campbell, R. F. Jurgens, R. B. Dyce, F. S. Harris, and G. H. Pettengill, "Radar Interferometric Observations of Venus at 70-Centimeter Wavelength," Science 170 (1970): 1090-1092; R. F. Jurgens, "Some Preliminary Results of the 70-cm Radar Studies of Venus," Radio Science 5 (1970): 435-442; and R. F. Jurgens, "A Study of the Average and Anomalous Radar Scattering from the Surface of Venus at 70 Cm Wavelength," PhD diss., Cornell University, June 1968, also published internally as CRSR Research Report no. 297 (Ithaca: CRSR, May 1968).


Carpenter dropped out of radar astronomy and pursued a teaching career, while Goldstein continued to explore Venus. The 1969 inferior conjunction of Venus provided an opportunity to use range-Doppler mapping. Goldstein combined the 1969 data with earlier data, then applied a mathematical method devised by fellow JPL employee Howard C. Rumsey, Jr., which involved the construction of a large matrix of range and Doppler values.

The mapping process divided the surface of Venus into small cells 1/2° square in latitude and longitude. A column vector (X) consisted of the unknown reflectivities of these cells, while a second column vector (S) contained all the processed data from 17 days of [139] observations. Already, Goldstein and Rumsey were dealing with a large amount of data; vector X had about 40,000 components, vector S about 120,000 components. They expressed the relationship between vectors S and X as the equation


AX = S,


in which A was a matrix whose components could be computed from known parameters and the motion of Venus and Earth. Matrix A consisted of 120,000 by 40,000 components.

As the authors wrote, "Obviously, we cannot compute every component of a matrix with over 109 entries." The matrix was "so big," Goldstein recalled, "that we couldn't even read it into the computer except one line at a time."57 Despite the difficulty of handling the gargantuan matrix, Goldstein produced a number of somewhat unambiguous images of Venus. Once the 1969 data had been converted into a range-Doppler map, in which each resolution cell represented an area on the planet's surface, Goldstein made cumulative maps by adding earlier data. The north and south areas of the cumulative maps were similar, but not identical; however, the images suffered serious flaws, including the "runway" strip running more or less along the planet's equator. Nonetheless, Goldstein succeeded in resolving Greek letter alpha for the first time on a map. It was a roundish feature, about 1,000 km across.58

Goldstein continued to map Venus with Rumsey's mathematical approach, adding data taken during the 1970 inferior conjunction to that acquired in 1969. The 1970 data were better, being less noisy, because the Deep Space Network had increased the transmitter power of the Goldstone Mars Station from 100 to 400 kilowatts. The total system noise temperature stood at a low 25 K. Regions Greek letter alpha and Greek letter beta remained the dominant features of the JPL radar map.59

Meanwhile, at Arecibo and Haystack, radar astronomers were creating Venus images with their own techniques. At Arecibo, Cornell University doctoral student Ray Jurgens, with support from an NSF Faculty Fellowship, undertook the analysis of radar data taken during the 1964 inferior conjunction. Dyce and Pettengill had made the radar observations to supply the Planetary Ephemeris Program data base, not to make a range-Doppler map of Venus.60

In correlating the radar data with the Cytherean surface, Jurgens abandoned his own zero degree meridian in favor of a modified version of Carpenter's coordinate system that incorporated the latest pole position and rotation rate supplied by Irwin Shapiro from the PEP. Consequently, Goldstein's Greek letter alpha and Carpenter's F were not at the zero meridian but closer to 5° longitude. Jurgens identified the features he found by latitude and longitude (e.g., 20°,-102°, in which "-" indicated South latitude or West longitude), then compared his features with those discovered by Goldstein and Carpenter.

Jurgens gave particular attention to Goldstein's Greek letter beta region (Carpenter's group B, C, D, to which Jurgens added E), and he managed to locate most of Carpenter's features. In addition, Jurgens spotted a new feature near Goldstein's Beta. Borrowing from Tommy Thompson's lunar radar mapping work, Jurgens interpreted the feature as a ring structure, specifically a crater, and argued that such a crater might be caused by meteoric impact. Jurgens admitted that "although the evidence for a ring structure is not as strong as one might desire, it at least raises the question of whether such structures would be expected on Venus."61 Indeed, it was one of the first attempts to relate radar observations and geological interpretation.

[140] Before completing his dissertation, Jurgens observed Venus during the 1967 inferior conjunction, when the Arecibo antenna had an improved receiver system, better data acquisition procedures, and a lower receiver noise temperature. Jurgens combined the 1967 data with additional observations made during the subsequent 1969 conjunction. In order to mitigate the north-south ambiguity problem, he compared observations made a few weeks apart, thereby taking advantage of the changing Doppler geometries between Earth and Venus.

Jurgens continued to explore the Greek letter beta region, in particular, as well as new areas of the planet's surface. On the urging of Tommy Gold, he named his features after scientists famous for their work in electromagnetism: Karl Friedrich Gauss (1777-1855), Heinrich Rudolph Hertz (1857-1894), Michael Faraday (1791-1867), and James Clerk Maxwell (1831-1879). Gauss and Hertz both corresponded strongly to Goldstein's Greek letter beta region. Faraday was Goldstein's Greek letter alpha. However, Maxwell, discovered during the 1967 conjunction, had no match among previous citings of Cytherean surface features. 62 It was an original and enduring contribution to Venus mapping.


Figure 20. Ray Jurgens discovered a new Venus surface feature, named Maxwell, from these range-Doppler images made at the Arecibo Observatory on 4 September 1967 during inferior conjunction.

Figure 20. Ray Jurgens discovered a new Venus surface feature, named Maxwell, from these range-Doppler images made at the Arecibo Observatory on 4 September 1967 during inferior conjunction. In A, the bright spot at the leading edge of the image is the subradar point, while the spot closest to the subradar point is the Beta region. Maxwell is the spot farther from the planet's leading edge. In B, the average scatter has been removed to show Beta and Maxwell better. (Courtesy of Ray Jurgens.)


[141] Investigators at Haystack Observatory also observed Venus during the 1967 conjunction, but they used a unique technique they pioneered called radar interferometry. It resolved the problem of north-south ambiguity in a superior fashion. An optical interferometer is an instrument for analyzing the light spectrum by studying patterns of interference, that is, how light waves interact with each other. Martin Ryle and other radio astronomers had been designing interferometers since the late 1950s. These radio interferometers used two or more radio telescopes arranged along a straight line (called the base line) and allowed astronomers to "synthesize" observations at higher resolutions than possible with a single antenna.63

The inventor of the radar interferometer was Alan E. E. Rogers, then an electrical engineering graduate student at MIT. MIT Prof. Alan H. Barrett was recruiting students to participate in his radio astronomy work on the newly discovered OH spectral line. Alan Rogers joined him and did his masters and doctoral theses on the OH line. As part of his doctoral thesis research, Rogers helped to develop a radio interferometer that linked the Millstone and Haystack radars.

After graduating and spending a year home in Africa, Rogers returned to Lincoln Laboratory, where he was hired to work in the radar group with Gordon Pettengill. Although trained as a radio astronomer, Rogers rapidly became absorbed in planetary radar work and proposed a radar interferometer to eliminate the problem of north-south ambiguity that was typical of range-Doppler mapping.64 This was not the first time that a radar astronomy technique derived from radio astronomy.

The X-band (7,840 MHz; 3.8 cm) radar interferometer linked the Haystack and Project Westford antennas, which are 1.2 km apart, in the so-called Hayford configuration. In the interferometry experiments, Haystack transmitted a continuous-wave signal to Venus, and both the Haystack and Westford antennas received. Technicians working under Dick Ingalls of Haystack reduced and analyzed the echoes to create a range-Doppler map. The size of the resolution cell on the planet's surface was about 150 km square.



Figure 21. One of the first range-Doppler image of Venus made with a radar interferometer, the Haystack and Westford antennas in tandem, in 1967.

Figure 21. One of the first range-Doppler image of Venus made with a radar interferometer, the Haystack and Westford antennas in tandem, in 1967. Not only are the Alpha and Beta regions discernible, but the complexity of Beta is revealed. (Courtesy of Alan E. E. Rogers.)


Next, Rogers and Ingalls combined the signals from the two antennas to obtain the fringe amplitude and phase for each range-Doppler cell. In an elaborate computer procedure, they rotated the fringe pattern so that the lines of constant phase were normal to the axis of apparent rotation of the planet. The lines of constant phase now were perpendicular to the slices of equal Doppler value. Although each pair of resolution cells [143] that exhibited north-south ambiguity had the same range and Doppler shift values, one could distinguish the north and south cells because they had opposite phases.65

One of the first applications of this radar interferometer was to the lunar work being carried out at Haystack for the NASA Manned Spacecraft Center by Stan Zisk. The lunar topographic maps that Zisk created with the Hayford interferometer were carried out under the name "Operation Haymoon" until December 1972, when the Apollo mission ended.66 Tommy Thompson carried out a similar interferometric study of the Moon using the 40-MHz (7.5-meter) radar at Arecibo.

Alan Rogers and Dick Ingalls also studied Venus during the 1967 inferior conjunction with the Hayford interferometer and identified eight surface regions. Just as each previous radar astronomer had invented his own nomenclature, they labeled features with Roman numerals and letters. The features of which they were certain became Haystack I through Haystack IV. The probable regions were Haystack A through Haystack D. Five of these eight regions corresponded to features already observed by either Goldstein or Carpenter. Haystack I appeared to be Goldstein's Greek letter alpha and Carpenter's F, while Haystack II matched Goldstein's Greek letter beta (Table 2). Jurgens' Arecibo results had not yet been published.67

Alan Rogers and Dick Ingalls then published a map of Venus showing the correlation of Haystack and JPL features in a 1969 issue of Science.68 The Greek letter beta region now appeared to be large and complex. The Hayford radar interferometer confirmed and extended the observations of Goldstein and Carpenter. With interferometer data taken during the 1969 and 1972 conjunctions using an instrument with a lower system noise temperature, Alan Rogers and Dick Ingalls refined their map of Venus; the data continued to indicate agreement among the Haystack and JPL features.69

The waxing tide of links between Lincoln Laboratory and Arecibo set in motion by the appointment of Gordon Pettengill as associate director of Arecibo facilitated the transplanting of radar interferometry to Arecibo. In fact, investigators at Arecibo built two additional antennas to study the Moon and Venus with the new technique. The lunar interferometer used the 40-MHz antenna, while the planetary radar interferometer used the 430-MHz antenna. NASA continued to underwrite Tommy Thompson's lunar radar work through a supplementary grant.70



Figure 22. Diagram of Venus surface features made with the Haystack-Westford interferometer.

Figure 22. Diagram of Venus surface features made with the Haystack-Westford interferometer. Features observed with the Haystack-Westford interferometer are indicated variously by capital letters, Roman numerals, and coordinate numbers. Goldstein's Alpha and Beta regions are indicated (Region Greek letter alpha and Region Greek letter beta), while the labels given by Carpenter are shown in parentheses. (Courtesy of Alan E. E. Rogers.)


Figure 23. The antenna built by Arecibo Observatory employee and radio amateur Sam Harris and located about 10 km from the main dish at Higuillales.

Figure 23. The antenna built by Arecibo Observatory employee and radio amateur Sam Harris and located about 10 km from the main dish at Higuillales. Harris and his antenna are a reminder of the importance self-taught engineers and radio amateurs have played in the design and construction of scientific instruments, particularly in the field of astronomy. (Courtesy of Ray Jurgens.)


Undertaking radar interferometric observations of Venus at Arecibo was Cornell graduate student Don Campbell. Campbell came to Cornell from Australia, his native country, where he had studied radio astronomy at the University of Sydney, though not through the agreement between the two universities.71 His observations of Venus in 1969 with the radar interferometer formed the basis of his doctoral thesis. Located about 10 km from the Arecibo Observatory at Higuillales near Los Caños, the auxiliary interferometer antenna was a square parabolic section, 30 meters by 30 meters (100 ft by 100 ft) with a movable offset feed that allowed tracking up to 10° from the zenith.72



Figure 24. Radar interferometric image of Venus made by Don Campbell for his 1971 doctoral dissertation, which was a study of Venus using the Arecibo Observatory and Higuillales antennas as a radar interferometer.

Figure 24. Radar interferometric image of Venus made by Don Campbell for his 1971 doctoral dissertation, which was a study of Venus using the Arecibo Observatory and Higuillales antennas as a radar interferometer. The resolution is about 150 km. The Alpha region can be seen in the lower right corner, and Beta Regio is visible in the upper left corner. (Courtesy of D. B. Campbell, Cornell University.)


As Don Campbell remembered, the original antenna was owned by Sam Harris, an Arecibo employee, who used it for his backyard amateur radio Moon bounces. Harris was a self-taught engineer well known in the "ham" community for his Moon-bounce work and had a column in the popular ham journal QST for many years. "He was a real character," Campbell reflected. "I always enjoyed working with him and in getting this interferometer to work over the year or so that it took."73

The 430-MHz radar interferometer went into operation in March 1969. Jurgens, Campbell, and Dyce, made interferometric observations of Venus between 20 March and 27 April 1969. Unfortunately, because the interferometer antenna was so small, the radar sensitivity fell sharply, and they achieved a surface resolution of only 300 km. The north-south ambiguity had been resolved, but at the loss of resolution. From the data, nonetheless, Campbell deduced that the Faraday region was the same as Goldstein's Greek letter alpha [147] and Carpenter's F. He concluded, "Despite the considerable advance that the radar interferometer represents over other methods in mapping the surface scattering of Venus at radio wavelengths, we still know very little about the actual nature of the surface."74 In other words, the images really said nothing about the planet's geology.

Campbell returned to Cornell, wrote his thesis, and graduated in July 1971. He then returned to Arecibo as a Research Associate employed by the NAIC. An improved line feed promised to improve observations during the next Venus inferior conjunction in 1972. Although delays in manufacturing the new analog-to-digital converters, as well as power outages, caused lost observing time, Campbell mapped Venus with the radar interferometer and achieved a resolution of about 100 km. "That was the last fling prior to the upgrade," Campbell recalled.75

Campbell also derived Venus topographical (relief or surface height) information from the 1972 data. The most notable result was the discovery of what appeared to be a mountainous zone located at a longitude of 100° and having a peak height of about 3 km. Although not at the same location as Jurgens's suspected crater, which still remained noted only in his dissertation, these mountains became the second clearly identified topographical feature on the surface of Venus, following a pioneering study by Smith and other Lincoln Laboratory and MIT investigators at Haystack published two years earlier.76

Dick Goldstein also observed Venus in 1972 with a radar interferometer that combined the Mars Station with a nearby 26-meter antenna. These were Goldstein's first Venus observations with an interferometer. Because he did not suffer the obstacles thrown at Don Campbell, Goldstein was able to update his large-scale, low-resolution map of Venus, which now had a resolution of 10-15 km. He also assembled his first altitude map. A gray scale of only five levels, with each level representing a set of altitude values, indicated the degree of relief. The map showed a large crater about 160 km in diameter about 36° West longitude and 2° South latitude. Goldstein estimated the height of the crater rim to be about 500 meters above the crater floor. This was the first distinguishable crater Goldstein found in his radar data; several years earlier, though, Ray Jurgens had identified a crater in his Cornell dissertation.77

By the 1972 inferior conjunction of Venus, the combination of range-Doppler mapping and radar interferometry was beginning to reveal a general overview of the planet's major surface features. Although Venus still looked like a strange fish bowl in radar images, lunar range-Doppler images looked more like photographs. These initial tentative steps, whatever their drawbacks, began to set in motion a shift in the planetary radar paradigm from astronomy to geology. Like the far more successful (because they looked like and had greater resolution than ground-based photographs) lunar radar images, Venus radar images showed that planetary radar astronomy could tell scientists useful information about distant surface formations. These images were not the only techniques radar astronomers had for describing planetary surface conditions. Coincidental with the gradual evolution of planetary radar toward these geological problems, NASA was turning from Apollo to planetary missions.



1. The works of Kuhn, which span over thirty years, have been summarized, explained, and analyzed in Paul Hoyningen-Huene, Reconstructing Scientific Revolutions: Thomas S. Kuhn's Philosophy of Science, trans. Alexander T. Levine (Chicago: University of Chicago Press, 1993). Especially relevant to the discussion here are pp. 134-135, 143-154, 169, 188-190 & 193-194.

2. Eshleman 9/5/94.

3. Hagfors and Campbell, "Mapping of Planetary Surfaces by Radar," Proceedings of the IEEE 61 (September 1973): 1219-1225, esp. 1224.

4. Kotelnikov, G. Ya. Guskov, Dubrovin, Dubinskii, Kislik, Korenberg, Minashin, Morozov, Nikitskiy, Petrov, G. A. Podoprigora, Rzhiga, A. V. Frantsesson, and Shakhovskoy, "Radar Observations of the Planet Mercury," Soviet Physics - Doklady 7 (1963): 1070-1072. Given the stated weakness of the Mercury echoes, as well as their difficulty in obtaining accurate and verifiable Venus results, the Soviet announcement of a detection of Mercury, a much farther radar target than Venus, raised doubts in the United States about the validity of the Soviet claims.

5. Carpenter and Goldstein, "Radar Observations of Mercury," Science 142 (1963): 381.

6. Pettengill and Dyce, "A Radar Determination of the Rotation of the Planet Mercury," Nature 206 (19 June 1965): 1240.

7. Dyce, Pettengill, and Shapiro, "Radar Determination of the Rotations of Venus and Mercury," The Astronomical Journal 72 (1967): 351-359.

8. Shapiro 30/9/93; Giuseppe Colombo, "Rotational Period of the Planet Mercury," Nature 208 (1965): 575.

9. Colombo and Shapiro, "The Rotation of the Planet Mercury," The Astrophysical Journal 145 (1966): 296-307. Earlier, it had appeared as an internal SAO publication: Colombo and Shapiro, The Rotation of the Planet Mercury, SAO special report no. 188 (Cambridge: SAO, 13 October 1965).

10. Shapiro 30/9/93; Peter Goldreich, "Tidal De-spin of Planets and Satellites," Nature 208 (1965): 375-376; Goldreich and Stanton Peale, "Resonant Spin States in the Solar System," Nature 209 (1966): 1078-1079; Goldreich, "Final Spin States of Planets and Satellites," The Astronomical Journal 71 (1966): 1-7; Goldreich and Peale, "Spin-Orbit Coupling in the Solar System," The Astronomical Journal 71 (1966): 425-438. Also, in a joint paper, Peale and Gold attempted to explain the rotational period of Mercury in terms of a solar tidal torque effect. Peale and Gold, "Rotation of the Planet Mercury," Nature 206 (1965): 1241-1242.

11. Shapiro 30/9/93; Counselman, "Spin-Orbit Resonance of Mercury," Ph.D. diss., MIT, February 1969. See also Counselman, "The Rotation of the Planet Mercury," Chapter 14, pp. 89-93 in R. G. Stern, ed., Review of NASA Sponsored Research at the Experimental Astronomy Laboratory (Cambridge: MIT, 1967).

12. Goldstein and Willard F. Gillmore, "Radar Observations of Mars," Science 141 (1963): 1171-1172.

13. Memorandum, O. Koksharova to I. Newlan, 9 January 1964, translation of Pravda article, microfilm 22-314, JPL Central Files. The article later appeared as Kotelnikov, Apraksin, Dubrovin, Kislik, Kuznetsov, Petrov, Rzhiga, Frantsesson, and Shakhovskoi, "Radar Observations of the Planet Jupiter," Soviet Physics-Doklady 9 (1964): 250-251.

14. Goldstein 14/9/93; Goldstein, "Radar Observations of Jupiter," Science 144 (1964): 842-843.

15. Dyce, Pettengill, and Sánchez, "Radar Observations of Mars and Jupiter at 70 cm," The Astronomical Journal 72 (1967): 771-777.

16. Goldstein, "Radar Observations of Icarus," Science 162 (1968): 903.

17. Dyce, "Attempted Detection of the Asteroid Icarus," in AIO, Research in Ionospheric Physics, Research Report RS 74 (Ithaca: CRSR, 31 July 1968), pp. 90-91.

18. "Weekly Reports, 5/13/68-8/11/69," 36/2/AC 135, MITA. The results appeared as Pettengill, Shapiro, Michael E. Ash, Ingalls, Louis P. Rainville, Smith, and Melvin L. Stone, "Radar Observations of Icarus," Icarus 10 (1969): 432-435.

19. Goldstein, "Icarus," pp. 903-904.

20. T. Gehrels, E. Roemer, R. C. Taylor, and B. H. Zellner, "Minor Planets and Related Objects: 4. Asteroid (1566) Icarus," The Astronomical Journal 75 (1970): 186-195; J. Veverka and W. Liller, "Observations of Icarus: 1968," Icarus 10 (1969): 441-444; Goldstein, "Radar Observations of Icarus," Icarus 10 (1969): 430-431.

21. "Editor's Introduction to: A Symposium on Icarus," Icarus 10 (1969): 429.

22. Tatarewicz, pp. 122-123.

23. Shapiro, Prediction of Ballistic Missile Trajectories from Radar Observations (New York: McGraw-Hill, 1958); Shapiro 4/5/94.

24. Campbell 9/12/93.

25. Campbell 7/12/93.

26. Shapiro 30/9/93.

27. Shapiro 30/9/93.

28. Shapiro 30/9/93; Shapiro 4/5/94; E. Myles Standish, Jr., telephone conversation, 20 May 1994; Paul Reichley, telephone conversation, 19 May 1994; Memorandum, Standish to R. Green, 10 May 1979, Jurgens materials.

29. Hine 12/3/93; Log books, Haystack Planetary Radar, HR-70-1, 9 December 1970 to 11 August 1971; HR-71-1, 16 August 1971 to 14 April 1972; HR-73-1, 27 June 1973 to 26 November 1973; and HR-73-2, 9 December 1970 to 11 August 1971, SEBRING. There is a lacuna in the log book records; observations made after 14 April 1972 and before 27 June 1973 are not represented.

30. Ash, Shapiro, and Smith, "Astronomical Constants and Planetary Ephemerides Deduced from Radar and Optical Observations," The Astronomical Journal 72 (1967): 338.

31. The section on Einstein's general theory of relatively draws loosely from Banesh Hoffmann, Relativity and its Roots (New York: W. H. Freeman and Company, 1983); Peter G. Bergmann, The Riddle of Gravitation, revised and updated (New York: Charles Scribner's Sons, 1987); and Mendel Sachs, Relativity in Our Time: From Physics to Human Relations (Bristol, PA: Taylor & Francis, 1993). See also Klaus Hentschel, "Einstein's Attitude towards Experiments: Testing Relativity Theory, 1907-1927," Studies in History and Philosophy of Science 23 (1992): 593-624.

32. Shapiro's recounting of the conception of the Fourth Test is included here, because the three-decade-long feud that resulted from it has become a part of the lore of radar astronomy. The sources for Shapiro's and Muhleman's versions of the story are oral histories conducted specifically for this history, namely Shapiro 1/10/93 and Muhleman 19/5/94. Paul Reichley, in a telephone conversation of 19 May 1994, refused any other comment than to state that he agreed with whatever Muhleman said.

33. Shapiro 1/10/93.

34. Shapiro 1/10/93.

35. Shapiro, "Fourth Test of General Relativity," Physical Review Letters 13 (28 December 1964): 789.

Although little noted at the time, Shapiro in his 1964 paper also pointed out that a possible change with atomic time of Newton's universal gravitational constant could be tested with radar observations of Mercury. Such a change was predicted by Paul Adrien Maurice Dirac in 1937 in his "large numbers hypothesis." Evidence for such a change is being actively sought still from monitoring orbits, as Shapiro suggested, because any such change would have profound effects on the evolution of the universe and the formation of structure within it.

36. C. Robert Wieser to Gen. B. A. Schriever, 31 May 1966, 13/56/AC 118, MITA.

37. See Shapiro, "Fourth Test," pp. 789-791; Muhleman and Reichley, "Effects of General Relativity on Planetary Radar Distance Measurements," in Supporting Research and Advanced Development, Space Programs Summary 37-29 (Pasadena: JPL, 31 October 1964), pp. 239-241. Although Muhleman's note had an earlier publication date, it was in an internal report with a tightly limited distribution, whereas Shapiro published in a widely distributed scientific journal. Paul Reichley, Muhleman's co-author, was a young college graduate recently hired at JPL and worked with Muhleman on occultation studies of radio signals. Reichley, telephone conversation, 19 May 1994; and Muhleman 19/5/94.

38. Shapiro, Effects of General Relativity on Interplanetary Time-Delay Measurements, Technical Report 368 (Lexington: Lincoln Laboratory, 18 December 1964), pp. 1-2; and Shapiro, "Testing General Relativity with Radar," Physical Review 145 (1966): 1005-1010.

39. Muhleman 19/5/94; Shapiro 1/10/93; Shapiro, Effects of General Relativity, p. 2.

40. See, for example, Peter G. Bergmann, The Riddle of Gravitation, rev. ed. (New York: Charles Scribner's Sons, 1987), p. 158.

41. Shapiro 1/10/93; Shapiro, Pettengill, Ash, Stone, Smith, Ingalls, and Brockelman, "Fourth Test of General Relativity: Preliminary Results," Physical Review Letters 20 (1968): 1265-1269.

42. Shapiro 1/10/93.

43. Roger Manasse, The Use of Radar Interferometer Measurements to Study Planets, Group Report 312-23 (Lexington: Lincoln Laboratory, March 1957).

44. Green 20/9/93.

45. See Ch. 1, note 69, and Ch. 3, note 14.

46. Green 20/9/93.

47. Green 20/9/93; Leadabrand, "Radar Astronomy Symposium Report," pp. 1111-1115. Earlier, a more complete exposition of the theory appeared as Green, A Summary of Detection Theory Notions in Radar Astronomy Terms, Group Report 34-84 (Lexington: Lincoln Laboratory, 18 January 1960). See, also, Ch. 3, note 22.

48. Pettengill, "Measurements of Lunar Reflectivity Using the Millstone Radar," Proceedings of the IRE 48 (May 1960): 933-934.

49. Pettengill and John C. Henry, "Enhancement of Radar Reflectivity Associated with the Lunar Crater Tycho," Journal of Geophysical Research 67 (1962): 4881-4885. Pettengill's co-author was an MIT electrical engineering graduate student who used the experience in writing his master's thesis. Henry, "An Automated Procedure for the Mapping of Extended Radio Sources," M.S. thesis, MIT, 1965.

50. Source for the arc measurement of Venus: Goldstein, "Radar Studies of Venus," in Audoin Dollfus, ed., Moon and Planets (Amsterdam: North-Holland Publishing Company, 1967), p. 127.

51. Thompson 29/11/94; Thompson and Dyce, "Mapping of Lunar Radar Reflectivity at 70 Centimeters," Journal of Geophysical Research 71 (1966): 4843-4853.

52. Thompson 29/11/94; Pettengill and Thompson, "A Radar Study of the Lunar Crater Tycho at 3.8-cm and 70-cm Wavelengths," Icarus 8 (1968): 457-471, esp. 464.

53. The research was conducted under NASA grant NGR-33-010-024. NEROC, Semiannual Report of the Haystack Observatory, 15 July 1972, p. ii. See, also, Ch. 4, note 15.

54. Goldstein, "Preliminary Venus Radar Results," Journal of Research of the National Bureau of Standards, Section D: Radio Science 69D (1965): 1623-1625; Goldstein, "Radar Studies of Venus," in Dollfus, Moon and Planets, pp. 126-131. This article also appeared as Goldstein, Radar Studies of Venus, Technical Report 32-1081 (Pasadena: JPL, 1967).

55. Goldstein and Shalhav Zohar, "Venus Map: A Detailed Look at the Feature Greek letter beta," Nature 219 (1968): 357-358; Goldstein, "A Radar View of the Surface of Venus," Proceedings of the American Philosophical Society 113 (June 1969): 224-228. Goldstein's co-author, Shalhav Zohar, was a fellow JPL employee who developed much of the software used in the experiment.

56. Carpenter, "Study of Venus by CW Radar: 1964 Results," The Astronomical Journal 71 (1966): 142-152, especially pp. 148-151.

57. Goldstein 14/9/93.

58. Goldstein and Rumsey, "A Radar Snapshot of Venus," Science 169 (1970): 974-977.

59. Goldstein and Rumsey, "A Radar Image of Venus," Icarus 17 (1972): 699-703.

60. Jurgens 23/5/94.

61. Jurgens, "A Study of the Average and Anomalous Radar Scattering," pp. 71 & 87-110.

62. Jurgens, "Some Preliminary Results of the 70-cm Radar Studies of Venus," Radio Science 5 (1970): 435-442; AIO, Research in Ionospheric Physics, Research Report RS 74 (Ithaca: CRSR, 31 July 1968), pp. 84-85.

63. See Bracewell, "Early Work on Imaging Theory," pp. 167-190 and Scheuer, "Aperture Synthesis at Cambridge," pp. 249-265 in Sullivan.

64. Rogers 5/5/94.

65. For a description of the radar interferometer, see Rogers, Hagfors, Brockelman, Ingalls, Levine, Pettengill, and Weinstein, A Radar Interferometer Study of Venus at 3.8 cm, Technical Report 444 (Lexington: Lincoln Laboratory, 14 February 1968).

66. Rogers 5/5/94; Documents in 44/2/AC 135; "Haystack Operations Summary, 8/11/69-5/18/70," 37/2/AC 135; "Funding Proposal, "Programs in Radio Astronomy at the Haystack Observatory," NSF, 10/1/72-9/30/73," 28/2/AC 135, MITA; NEROC, Semiannual Report of the Haystack Observatory, 15 July 1972, p. ii; NEROC, Final Progress Report Radar Studies of the Planets, 29 August 1974, p. 1. A number of techniques for extracting lunar topography from interferometric data were devised. Delay-Doppler stereoscopy was developed by Irwin Shapiro and independently by Thomas Thompson and Stan Zisk. Another technique, called delay-Doppler interferometry, was suggested by Shapiro and developed by Zisk and Rogers; Thompson pointed out the strength of the Hayford interferometer for this application. Shapiro, Zisk, Rogers, Slade, and Thompson, "Lunar Topography: Global Determination by Radar," Science 178 (1972): 939-948, esp. notes 19 and 21, p. 948.

67. Thompson, "Map of Lunar Radar Reflectivity at 7.5-m Wavelength," Icarus 13 (1970): 363-370.

68. Brockelman, Evans, Ingalls, Levine, and Pettengill, Reflection Properties of Venus at 3.8 cm, Report 456 (Lexington: Lincoln Laboratory, 1968), especially pp. 34-35, 44, & 49-50; Rogers and Ingalls, "Venus: Mapping the Surface Reflectivity by Radar Interferometry," Science 165 (1969): 797-799.

69. Rogers and Ingalls, "Radar Mapping of Venus with Interferometric Resolution of the Range-Doppler Ambiguity," Radio Science 5 (1970): 425-433; Rogers, Ingalls, and Pettengill, "Radar Map of Venus at 3.8 cm Wavelength," Icarus 21 (1974): 237-241.

70. AIO, Research in Ionospheric Physics, Research Report RS 75 (Ithaca: CRSR, 31 March 1969), pp. 2 & 11-12; Annual Summary Report, Center for Radiophysics and Space Research, July 1, 1968-June 30, 1969, 30 June 1969, p. 4.

71. Campbell 7/12/93.

72. Donald B. Campbell, "Radar Interferometric Observations of Venus," Ph.D. diss., Cornell, July 1971; AIO, Research in Ionospheric Physics, Research Report RS 75 (Ithaca: CRSR, 31 March 1969), pp. 12-13; Campbell, Jurgens, Dyce, F. Sam Harris, and Pettengill, "Radar Interferometric Observations of Venus at 70-Centimeter Wavelength," Science 170 (1970): 1090-1092.

73. Campbell 7/12/93.

74. Campbell 7/12/93; AIO, Research in Ionospheric Physics, Research Report RS 75 (Ithaca: CRSR, 31 March 1969), pp. 12-13; Ibid., Research Report RS 76 (Ithaca: CRSR, 30 September 1969), p. 23; Campbell, Jurgens, Dyce, Harris, and Pettengill, "Radar Interferometric Observations," pp. 1090-1092.

75. Campbell 7/12/93; NAIC QR Q2/1972, pp. 3-4, and Q3/1973, pp. 3-4.

76. Campbell, Dyce, Ingalls, Pettengill, and Shapiro, "Venus: Topography Revealed by Radar Data," Science 175 (1972): 514-516. Smith, Ingalls, Shapiro, and Ash, "Surface-Height Variations on Venus and Mercury," Radio Science 5 (1970): 411-423, presented an earlier topographical study of Venus made from Haystack data taken over a period of years. That study was confined to the planet's equator and found a 2-km feature. It was remarkable for the variety of radar techniques used, as well as for its discovery of the first topographical feature on Venus.

77. Rumsey, Morris, R. Green, and Goldstein, "A Radar Brightness and Altitude Image of a Portion of Venus," Icarus 23 (1974): 1-7; Jurgens, "A Study of the Average and Anomalous Radar Scattering," pp. 87-110.