Doc Ewen's prologue to Paul Goldsmith's recollections

:

The SWAS program was blessed with the right combination of a talented Engineering Team and Science Team. The Engineering Team designed, assembled, tested and launched the SWAS into orbit. The Science Team conceived the project, pursued the funding and used the unique power of the telescope to obtain and analyze data that provided new insight concerning the deep space distribution of molecular species and processes associated with the building blocks of carbon, hydrogen and oxygen.

Members of the SWAS Engineering Team whose unique talents made the receiver development possible included:

A complete list of all members of the SWAS Receiver Engineering Team is presented in Slide number 46 of the slides on Microwave and Millimeter Wave Applications in the 1970s and 1980s.[***insert link when slide page completed***]

A PERSONAL PERSPECTIVE by Paul Goldsmith follows. Paul is a member of the SWAS Science Team and an early participant in the definition of the project.




The Submillimeter Wave Satellite (SWAS)

A Personal Perspective

Paul F. Goldsmith
Department of Astronomy, Cornell University


Origins and Start of the SWAS Project

The SWAS project had its origins in conversations between Martin Harwit of Cornell University and myself during the mid-1980s, when I was on the faculty of the Department of Physics and Astronomy at the University of Massachusetts, Amherst. We discussed creating a NASA proposal for a small satellite to observe key spectral lines in the submillimeter wavelength range. Martin was well aware of the importance of interstellar water vapor for cooling interstellar clouds and allowing them to contract and form stars, and I had a long-standing interest in molecular oxygen, so we included these lines in our range of wavelengths. We created a proposal to study the idea in more detail, submitted it, but NASA turned it down, in part because the agency was about to release an announcement of opportunity for its "Small Explorer" or SMEX program - a program that seemed ideal for a small-scale project such as this one.

In order to develop a much more detailed proposal a larger team was necessary. Gary Melnick at the Smithsonian Astrophysical Observatory had been involved in a variety of efforts to detect the far-infrared emission from water in the interstellar medium, as well as modeling its emission. With this experience, and having become convinced that a space mission was the best route for this experiment, Melnick became deeply involved in the new effort and got others at SAO interested in SWAS science involved as well. Since SAO had considerable experience in equipment development and project management, as well as having several individuals interested in SWAS science, it was a natural choice to lead the new effort, and Melnick was ideally suited to be the Principal Investigator for the proposal.

Ball Aerospace was brought in as the prime contractor for the payload, and the Millitech Corporation for design and construction of the low noise submillimeter "front end" receiver. Ball had extensive experience in all aspects of spacecraft design, having worked relatively recently on the IRAS (Infra Red Astronomy Satellite) project. Millitech was in a very strong position to build a practical, low-power submillimeter front end as a consequence of its successful completion of a Small Business Innovation Research (SBIR) contract with NASA/JPL on just this topic. NASA's Goddard Spaceflight Center provided the spacecraft itself, including the solar cells, power, computer, and telemetry systems. Additional members of the science team were from the Smithsonian Astrophysical Observatory, the University of California, Berkeley, the University of Massachusetts, Amherst, the University of Cologne (Germany), and the NASA/Ames Research Center.

The original SWAS proposal included most of the major concepts that ended up as part of the flight hardware. These included the use of second harmonic mixers, pumped by frequency-multiplied Gunn oscillators. The harmonic mixer is an old standby for cases in which a local oscillator at the desired high frequency is not available, and in fact I had used one in my PhD research on the then "high" frequency of 230 GHz. This mixer design, along with the frequency multipliers, had been fully developed by Neal Erickson, and had reached the point where noise temperatures of ~2500 K (DSB) at a physical temperature of 150 K could be anticipated for frequencies of ~500 GHz. The complement of spectral lines to be observed, those of neutral carbon, molecular oxygen, water, and the 13C isotope of carbon monoxide was unchanged from start to finish of the SWAS project, although the astronomical results turned out to be far different from those anticipated in the original proposal!

The initial SWAS proposal was submitted to NASA in September 1988, with Gary Melnick as the PI and the Smithsonian Astrophysical Observatory as the proposing institution. The proposal included subcontracts with the University of Massachusetts as well as with other institutions and companies. At this point, the anticipated launch vehicle was the SCOUT rocket, a veteran of many successful launches from a platform floating in the Indian Ocean just east of Kenya. NASA selected SWAS for development in 1989, but instead of moving directly to design and fabrication of flight hardware, initiated a series of study programs. Part of the reason for these studies was the need to verify the readiness of all critical technology. Many of the key components - the mixers, local oscillators, and IF system - were to be very similar to units that had already been built and demonstrated to work satisfactorily on orbit. However, SWAS's plan to have the front end cooled to ~160 K, while nothing extreme for ground-based systems, would represent a new challenge for spaceborne radiometers. Acousto-optical spectrometers, while widely used in radio astronomy facilities on the ground, had never been flown, so that this component would have to be carefully tested. Additionally, SWAS was certainly more complex than the two other missions selected for the first round of SMEX projects, and so the slower pace gave us more time to thoroughly study all aspects of our mission.

This early phase of the SWAS project also saw major changes in NASA's perspective on the mission. The SCOUT rocket launch from Kenya into an equatorial Earth orbit was changed to a launch from Wallops Island, Virginia and a necessarily inclined orbit. Subsequently, an even more significant modification occurred with a change in the launch vehicle. The new rocket, the PEGASUS, was being developed by Orbital Sciences Inc., and had the unusual property of being carried in the belly of an airplane to an elevation of about 10 km. When the aircraft attained the proper altitude, speed, and direction, it released the PEGASUS, whose rocket engine ignited a few seconds later, carrying the satellite into orbit. The aircraft's altitude and speed significantly increased the launch capability of the rocket. The change of launch vehicle led to an increase in the volume and mass of SWAS, which allowed us to change the antenna dimensions from the 55-cm circular aperture initially proposed to the 54 x 68 cm ellipsoidal antenna that was launched. This not only provided increased collecting area but also led to a very high-efficiency coupling to the radiation patterns of the conical feed horns that were employed. The increased mass that could be taken to orbit greatly enhanced the flexibility in the design allowing larger reaction wheels to nod the whole spacecraft, and avoided the need to use expensive lightweight materials such as SiC.

A variety of other issues produced further delays, during which NASA replaced the PEGASUS was with the PEGASUS XL. Following some early launch vehicle failures that resulted in additional schedule slippage, the PEGASUS XL placed SWAS almost exactly in the desired orbit on December 5, 1998. In fact, the SWAS instrument was ready in 1995, and the additional delay was entirely due to waiting for a launch. The final interval of just over 10 years between the proposal submission and the satellite launch did not conform to the concept of the SMEX program, which was expected to support relatively "quick" space projects, but the satellite’s nearly perfect on-orbit performance has nicely compensated for the extended waiting period.

Scientific Goals of SWAS

SWAS's basic scientific goals were the study of interstellar chemistry and star formation. Atomic and molecular line radiation plays the crucial role in cooling any dense molecular cloud in interstellar space, thus allowing it to contract to form a star. A key coolant is water (H2O), which had previously been detected in interstellar space, but only by means of transitions with peculiar properties that cause water molecules to act as natural amplifiers or masers. The 557 GHz ground state transition observed by SWAS is easily excited under the physical conditions expected in dense interstellar clouds (temperatures < 50 K and densities ~ 1 million hydrogen molecules per cubic cm). The intensity of this transition would provide an important tracer of cooling rate in interstellar clouds. This frequency is unmeasurable from the Earth's surface because of the absorption by water vapor in our atmosphere.

Molecular oxygen (O2), like water, was expected, on the basis of existing astrochemical models, to be highly abundant in interstellar clouds. While the weakness of its magnetic dipole transitions meant that oxygen should not be a significant coolant, oxygen molecules were expected to provide a valuable tracer of the chemistry and physical conditions in interstellar clouds. As with water, absorption in Earth’s atmosphere makes molecular oxygen unobservable even from the highest mountain tops or with airborne telescopes. The study of these important molecular species in interstellar clouds therefore requires a space mission to carry the observing platform completely above the atmosphere. This fact provided the basic justification for the SWAS mission.

Atomic carbon was an unexpected entry in our observing list since it is not a molecule, but this species had been previously found to be surprisingly abundant in molecular clouds. Observations of the 492 GHz fine-structure transition of atomic carbon (CI) are possible from the highest-altitude observatories on Earth under the driest conditions (because atmospheric absorption arises primarily from water vapor), but these conditions also make accurate calibration problematic. The presence of CI in interstellar clouds offered a tantalizing hint that we were missing some important aspect of the structure or evolutionary history of these regions, and SWAS was expected to carry out large-scale studies that would dramatically improve our understanding of how and why this atomic species remains abundant in an almost totally molecular environment.

Carbon monoxide (CO) is well established as a good tracer of the molecular component of the interstellar medium. The carbon monoxide spectral line that SWAS can observe is the J = 5-4 rotational transition of 13CO. This is likely to be strong only in warm and dense interstellar regions, which makes it a highly selective probe of the regions of clouds heated by already-formed stars. This high transition is completely unobservable from the ground due to absorption by terrestrial water vapor.

SWAS Testing and Early Operation

The components of the antenna, radiometer, and spectrometer were all put through intensive tests at different stages of the design and construction phases of the SWAS project. Tests through multiple thermal cycles of the front-end components were judged essential to guarantee reliable cold operation on orbit, although in fact it was never anticipated that these components would ever warm after their post-launch cool down. This was a difficult challenge that was met by engineers and scientists at Millitech.

A significant problem in achieving complete system testing arose in making measurements to verify the performance of the antenna and feed subsystem. The difficulty here is that atmospheric absorption is so strong that by the time that you move a source into the far field of the antenna (distance ~ 2D2 / λ ~ 1.5 km for SWAS), the signal will be heavily absorbed, so traditional antenna pattern measurements will not be reliable. The search for an alternative approach led to the adoption of the near-field technique, in which the amplitude and phase in a plane near the antenna aperture are measured and the far-field beam pattern can then be obtained by using a Fourier transform. But to make such a system work properly at the ~500 GHz frequency of SWAS required overcoming a variety of technical challenges. SAO, Millitech Corporation and Near Field Systems Inc. (NSI) together developed a near-field test range with a total accuracy of 15 microns rms which incorporated several novel ideas (see the paper by D. Slater, NSI, 1994 Antenna Measurement Techniques Association Conference***full cite***). The measurements showed that the SWAS antenna and feeds had been correctly designed and fabricated, and just as important, had also been properly aligned. Thus, while not revealing any problems, these measurements greatly enhanced our confidence that SWAS was not going to suffer from an optical problem such as the one that had been found in 1990 to be plaguing the Hubble Space Telescope.

Another major test verified that the frequencies of all the oscillators in the SWAS receivers were correct, and that the antenna, feed, front end, IF, and spectrometer were all working properly. To do this, we simulated an astronomical signal by means of a gas cell. This was an evacuated glass cylinder which could be placed in front of the antenna, and into which a small amount of the desired molecular gas could be introduced. By giving the gas in the cell a temperature different from that of a blackbody load placed on the far side of the cell (as seen by the receiver), a signal of known intensity and frequency could be produced. Using H2O and H2CO (formaldehyde, which has a transitions close in frequency to those of the SWAS target spectral lines), we verified that we could detect signals much like those expected from interstellar clouds.

SWAS has a passively cooled front end: The input optics, feed horns, mixers, and first IF amplifiers are radiatively cooled. The cooling efficiency is increased by the use of Winston cone radiators seen in Slide 39 in Section 3 [***check reference and insert link***] of Doc Ewen's collection. Thus, after launch, once the satellite had been verified to have achieved the desired orbit, it was necessary to wait for the radiative cooling, which could begin only when the spacecraft was pointing towards "cold space," to take effect. We anticipated that it would take several days for the satellite to cool from its initial ambient temperature to the final ~160 K temperature determined by balance between heat load and radiative cooling. As the front end approached this temperature, it was possible to start observations. The pointing of SWAS's submillimeter beam was achieved by orienting the entire spacecraft. This task, which must be done with an accuracy of a fraction of the ~4' beam width, is carried out by a combination of gyroscopes and torque wheels, controlled by an optical star tracker.

Following the launch by the PEGASUS XL rocket, which did take place almost exactly according to plan, telemetry contact with SWAS was established. The figure below shows (from left to right) Harold Reitsema (Ball Aerospace), Gary Melnick, Christine Hemp (a poet interested in space exploration) and Ron Snell in the flight control room at Vandenberg Air Force Base in California, shortly after SWAS was launched.


[In the flight control room after SWAS launch]

Credit: Photo courtesy of Paul Goldsmith.


An initial scare following the launch was that the front end electronics seemed to be getting too cold! While this might in principle, have resulted in lower noise, it was dangerous because critical components had not been tested to temperatures as low as were being reached without any power being dissipated in the radiometer. To keep things from getting too cold, Gary Melnick requested that the front end electronics be turned on earlier than had been scheduled, and after this was done, the temperature stabilized at a reasonable value.

Once it had been verified that SWAS was in the desired orbit and the front end was operating, an important step in making SWAS operational was to determine the direction of the submillimeter beam relative to that of the star tracker. This was achieved by observations of the planet Jupiter, which provides a strong submillimeter source (radiating like a black body at ~165 K) as well as being easily optically visible. Initial scans of the satellite across Jupiter's expected position indicated no signal, provoking considerable anxiety on the part of the science team. The radiometers seemed to be working; they had close to the expected noise temperature, and the star tracker and pointing system were behaving properly, so what was the problem? Repeated efforts to locate the submillimeter beam prompted larger and larger scans around the nominal direction.

Finally, to everyone's enormous relief, a clear signal from Jupiter was identified. The explanation for the earlier non-detection was that the offset between the submillimeter antenna beam and the star tracker boresight differed from that calculated before launch. The offset was only a fraction of a degree, which had no adverse effect on long-term operations, but it did get create a high degree of anxiety at the outset of the mission. But the offset, once identified, turned out to be extremely stable, and the whole pointing system worked exceptionally well with the result that in routine operation the spacecraft could be pointed at astronomical objects with an rms accuracy of ~5 arcseconds.

The first astronomical spectrum was soon obtained and verified proper operation of the entire receiver system (as well as the software). Additional observations of Jupiter and Mars showed that the aperture efficiency of the telescope was a very satisfactory 66%. SWAS then started an extremely productive career during which it observed almost continuously for a period of six years. The performance of the SWAS instrument on the ground and on orbit is discussed in detail by Tolls et al. (ApJ Suppl., 152, 2004 May, 137). From the very first spectra obtained with SWAS, it was evident that emission from water, CI and 13CO could be fairly easily detected in a variety of interstellar clouds. During the first few months of operation, the science team made a number of large-scale maps of clouds delineating the distribution of these species. Maps of the Orion Molecular Cloud are shown in the accompanying figures. It surprised some to see how extensive the distribution of water was, given that it is difficult to excite this molecule (in terms of the density of collision partners), compared with the other species observed by SWAS, or, in fact, by any other interstellar molecules. So while the water emission is not as extended as those of CI or 13CO, it is proving a challenge to understand exactly how its emission can be seen over such large areas in Orion and other molecular clouds.


[Maps of Orion molecular cloud]


The left panel shows emission in the 557 GHz H2O line integrated over all velocities and the middle panel shows emission integrated over velocities between 1 and 10 km s-1. The contours are scaled to the peak values, which are different. Comparison of the two panels shows that the broad emission, which dominates the left panel, is concentrated at the position of the young, embedded source, IRC2, which is driving a molecular outflow, while the narrow emission from the quiescent cloud is seen in the middle panel to be more extended, especially to the north of IRC2. The difference between outflow and quiescent cloud is very clear in the spectra shown at the right of the following figure. The right panel here shows the integrated emission from the J = 1-0 transition of 13CO which delineates the quiescent cloud emission (Snell et al. 2000, ApJ, 539, L93).

What soon became clear, however, was that nowhere, even in the largest, densest clouds, was water emission anywhere near as intense as had been expected. As the data continued to arrive, and cross checks confirmed the calibration accuracy of the system, it grew ever more evident that the abundance of water was far less than had been predicted by the models of interstellar chemistry.


[3 views of Orion molecular cloud]

Credit: images courtesy of Paul Goldsmith.


Left - Three views of the Orion molecular cloud, in 850 micron dust continuum radiation, the CI fine structure transition, and the 13CO 5-4 transition. The latter two maps, obtained with SWAS, are among the largest maps of submillimeter tracers. Right - A visual image of the Orion nebula, located at the center of the Orion molecular cloud, and spectra of emission from 13CO, CI, H2O, and at upper left, H218O.

During the SWAS development program, we had decided to enable observation of the oxygen-18 variant of water, which has only 0.2 percent of the abundance of water with oxygen’s normal isotope, oxygen-16. This decision arose because we were concerned that the expected large abundance of water would produce a highly saturated spectral line. This in turn would make it almost impossible to determine the water abundance. In this situation, the less abundant variant would be relatively easy to observe and vital to determine how much water was present. But what we were finding was that emission from "normal" water is invariably quite weak, and that from H218O extremely so. This suggested that there was much less water present in the interstellar gas than we had expected. By integrating for many tens of hours, SWAS was finally able to detect signals from the isotopic variant, which confirmed the very low abundance of water overall - at least a factor of 100 below that anticipated. These long integrations also confirmed that SWAS provided one of the most stable radiometric systems ever operated, with output fluctuations decreasing as expected as a function of time up to the longest integration employed, of almost 400 hours duration!

The observations of molecular oxygen were even more surprising in that in no source was a convincing detection of this molecule made. The upper limit to the fractional abundance of O2 (relative to H2) was again about two orders of magnitude below that expected from the standard models. The results from the first nine months of SWAS operation were published in a special issue of The Astrophysical Journal Letters (Volume 539, 2000 August 20). This included a description of the SWAS instrument and its performance, a variety of observations of interstellar clouds in all of the four SWAS tracers, together with measurement of water in the atmospheres of Mars, Jupiter, and Saturn. The fortuitous appearance of comet C/1999 H1 (Lee) permitted us to make the first direct determination of the rate of production of water in a comet, which was also reported in the special issue.

Verifying SWAS Performance and A Surprise

The fact that no clear signal from O2 had been detected from any source gave rise to some concern that there was something wrong with the system. Could there be a frequency-selective filter cutting out response at the 487.25 GHz frequency of the O2 transition? Or was the frequency possibly incorrect? Fortunately, an accidental glitch in SWAS operations answered this question. It originated in a basic safety feature built into the spacecraft's operating system: If the computer decides there is a system anomaly, or does not receive instructions where to point, it automatically stows the satellite in a "safe hold" position looking away from the Earth. This direction is important because it keeps the thermal radiators mentioned above pointing into cold space and thus maintains the thermal equilibrium of the front end system.

At one point during 2001, a mix-up prevented the observing command sequence being uploaded from one of the two grounds stations that could be used, to SWAS, which normally took place once per day. Thus, when the file of observing commands had been completely executed, SWAS properly started to reorient itself to point in the safe hold direction. However, the software did not specify that in moving to the safe hold direction, SWAS had to avoid the Earth. As it turned out, while executing the slew maneuver, the antenna pointed towards the Earth and neatly scanned through the Earth's limb. Atmospheric limb scanning is a technique widely used in remote sensing experiments wishing to study the atmosphere, since different lines of sight through the limb probe different altitudes in the atmosphere.

SWAS was not intended to observe the Earth, and in fact the front end did start to warm up during this maneuver, but only by a few K. The system soon cooled back down to its equilibrium temperature with no damage at all. But what did happen is that while looking at the Earth's limb, exquisite spectra of emission from the oxygen transition of interest were obtained. We knew this line should be there - it is this oxygen that requires that SWAS be in space - and we could calculate its expected intensity. This agreed almost perfectly with what was observed.

[SWAS 487 GHz O2 spectra]


The figure shows SWAS 487 GHz O2 spectra from two different tangent heights (minimum altitude in the atmosphere reached by beam from SWAS) which reveal not only the oxygen line but also emission from two transitions of ozone (O3). These again had the expected intensities and line shapes, confirming that the SWAS antenna, radiometer, and spectrometer were all performing exactly as expected, and that there was no instrumental reason why we should not be able to see the O2 emission from interstellar clouds if it were indeed present. The terrestrial O2 spectrum was published in 2002 (Goldsmith et al., ApJ, 576, 814-831, 2002), as part of an analysis of a tentative detection of very weak molecular-oxygen emission from a moderate velocity molecular outflow in the Rho Ophiuchi cloud. The astronomical detection was interpreted as resulting from oxygen released from the grain surfaces by the shock having a chance to form O2 in the gas phase which could be relatively abundant for a brief period. This detection has not been confirmed, but whether or not it proves to be real, the results indicate an extremely low abundance of molecular oxygen in the quiescent cloud material, consistent with earlier results.

In one case SWAS found the very surprising "opposite" result - far more water than expected. In this case, the target was the evolved carbon star IRC+10216. This star has a few times the mass of our sun, and has evolved after consuming nearly all its hydrogen as nuclear fuel. During its post-main-sequence evolution, the star has expanded so much that if it replaced our sun, it would extend out to the radius of the planet Jupiter. During its evolution, IRC+10216 has converted a substantial fraction of its mass to carbon, so this atomic species is now far more common than the usually more abundant oxygen. This means that the chemistry in this stellar atmosphere will be characteristic of material with C/O > 1, and in fact will in some ways resemble interstellar matter in regions where oxygen has been depleted by entering the surfaces of interstellar dust grains. Again, we would not expect to find much water vapor, since all the available oxygen will be locked into carbon monoxide.

In any case, the SWAS team decided to do a long integration on IRC+10216, and was quite startled to see a hint of emission from water, which upon further integration emerged as a very clear detection. Detection of water in the atmosphere of a carbon star was regarded as a stunning surprise by most radio astronomers and astrochemists. Assuming that SWAS did detect water in IRC+10216, we must ask: Where could the water be coming from, if the stellar atmosphere on its own would have almost none?

The answer is not yet definitive, but the most likely explanation is that the water is being vaporized from the surfaces of icy bodies orbiting the star. While we do not know for sure that IRC+10216 possesses a Kuiper Belt of such bodies like the sun's, it appears plausible that it does. The increased heating from the swollen star is now heating the surfaces of all of these icy objects, and the quantity of water they are expected to release into the gas phase fits quite well with the amount of water measured by SWAS. This scenario, invoked by G. Melnick et al. (Nature, 412, 160M, 2001) to explain the SWAS observations, had been explored theoretically a decade earlier, but it is very difficult to be certain that there is no other explanation for the water seen in IRC+10216.

Additional support for the icy body origin of the water has been obtained by detection of formaldehyde in this source, which is a relatively abundant species in material vaporized from solar system comets (Ford et al., ApJ, 614, 990, 2004). While not definitively identifying the origin of the water, detection of OH in the atmosphere of this star using the Arecibo radio telescope by Ford et al. (ApJ, 589, 430, 2003) is consistent with the quantity of this radical expected from photodissociation of water by ultraviolet radiation. Future observations of a wide variety of species with higher angular resolution will be necessary to fully understand the complex chemistry that results when the outflowing atmosphere of the evolved star combines with new ingredients from objects engulfed by the giant star.

Astrophysical Significance of SWAS Results

What is the significance of the low abundance of water and oxygen in interstellar clouds? While this question has not yet fully been answered, the most likely scenario is that molecules are colliding with the dust grains known to exist in interstellar clouds and sticking to their surfaces. In particular, if oxygen atoms do this, they can then interact with hydrogen atoms also adsorbed on the grain surface to form water molecules. However, this water may remain frozen on the grain surface. This scenario is supported by the infrared detection of water ice in the spectrum of stars whose light shines through dense interstellar clouds. The signatures of water and other molecules "frozen" on grain surfaces had been studied for some time, particularly by infrared astronomers. One interesting result of the process of molecules sticking to grains - called depletion - is that the grains no longer possess pristine surfaces of silicates or carbon, but instead have mantles of water and carbon monoxide, mixed together with other species.

But there is another result, which is plausibly even more important. If oxygen atoms stick to grains, the gas phase is deprived of these atoms. This increases the gas-phase carbon-to-oxygen ratio, since the carbon atoms are not expected to stick to dust grains. In this situation, effectively all the remaining oxygen atoms will combine with carbon to make carbon monoxide, which is the most abundant molecule after H2 in these regions. This enormously reduces the amount of oxygen available for forming O2 and H2O. While some aspects of the interlocking of gas-phase and grain-surface chemistry had been suspected before, the SWAS results brought this relationship into the forefront of astrochemical modeling. Even today the ramifications of this gas-grain interaction are being further explored, in part because it produces a major impact on the enhancement of deuterium in particular types of molecules in which this heavy isotope replaces a hydrogen atom. It also reduces the gas cooling rate and has other important astrophysical effects.

I want to take this opportunity to acknowledge the terrific support throughout the SWAS project provided by many individuals at Millitech Corporation, Ball Aerospace, NASA Goddard Spaceflight Center, and the institutions with which members of the SWAS team are associated. Of course, my colleagues on the SWAS science team spent an enormous effort over a decade and a half to ensure its technical and scientific success, and I thank them all for their unstinting generosity. The final figure shows members of the SWAS team in Cambridge Massachusetts following one of our final science meetings.

[SWAS team]

From left to right in top row: David Hollenbach, Rene Plume, Martin Harwitt; next row: Ron Snell, Ralph Paganetti; next row: Frank Bensch, Di Li, Paul Goldsmith, Mark Gurwell; front row: Gary Melnick, Ted Bergin, John Howe, David Neufeld. Credit: Photo courtesy of Paul Goldsmith.