Experimental Astronomy

, 25:125 | Cite as

History of infrared telescopes and astronomy

Review Article


The first attempts to measure the infrared outputs of stars preceded by nearly a century the permanent establishment of infrared astronomy as an important aspect of the field. There were a number of false starts in that century, significant efforts that had little impact on the astronomical community at large. Why did these efforts fizzle out? What was different in the start that did not fizzle, in the 1960s? I suggest that the most important advances were the success of radio astronomy in demonstrating interesting phenomena outside of the optical regime, and the establishment virtually simultaneously in the United States of a number of research groups that could support each other and compete against one another in their approach to infrared astronomy.


Infrared astronomy History of astronomy 

1 Introduction

The infrared spectral range has moved to the center of our goals for astronomy. Future facilities including Herschel, the James Webb Space Telescope (JWST), the Space Infrared Telescope for Cosmology and Astrophysics (SPICA), the Stratospheric Observatory for Infrared Astronomy (SOFIA), and 30-m-class groundbased telescopes will finally let us reach the potential of this field. However, infrared astronomy began fitfully and continued that way until less than 50 years ago, when it suddenly began to grow rapidly.

In this article, I discuss this early, fitful development of galactic and extragalactic infrared astronomy, and then briefly touch on the beginnings of modern infrared astronomy in the 1960s and the rapid growth that ensued. In this way, I can complement other material relevant to the history of this field. For example, Lequeux (2008, private communication) describes the very early stages of infrared astronomy; Connes [15] describes the development of Fourier Transform infrared spectroscopy; Sinton [82] puts emphasis on the development of planetary infrared astronomy; Price [72] describes infrared surveys, including a number that never received broad recognition; Jennings [37] and Storey [86] emphasize the growth of infrared astronomy in England and Australia, respectively; Dolci [17] discusses the history of airborne astronomy in general, which includes its major role in the infrared; Rieke [76] describes the development of the Spitzer Space Telescope; and Low et al. [50] describe the rapid emergence of infrared astronomy in the 1960s. Although a number of holes remain (e.g., for most infrared space missions), the combination of these articles starts to yield a reasonably complete description of the emergence of the field.

2 False starts

2.1 Nineteenth century

Infrared astronomy had a faltering beginning. Huggins [34] built a relatively simple thermopile with a galvanometer readout, for an 8-in. refractor. Similar instrumentation had been applied by Luigi Magrini to detect the solar corona during the eclipse of 1836 and by the fourth Earl of Rosse (son of the builder of the 72-in. “Leviathan” Telescope) to study the thermal radiation of the Moon [74, 79]. Despite his modest instrument, Huggins reported detections of Sirius, Pollux, Regulus, and Arcturus. Stone [85] improved the reliability of thermal measurements by using a matched pair of thermocouples at the telescope focus, to compensate both for thermal drifts in the instrument and for fluctuations in the emission of the sky (a measure also adopted by Lord Rosse). Stone reported detections of Arcturus and Vega. Only Lord Rosse seems to have made an effort to demonstrate that he was observing predominantly in the infrared, by putting a piece of glass in the beam and observing the resulting reduction in the signal. The others evidently assumed this result, given the broad response of their thermopile detectors.

The American inventor Edison excelled at discovering new applications for capsules of finely divided carbon powder. One example was an infrared detector consisting of a pellet of this material with a galvanometer to record its resistance as it reacted to changes in temperature. As a form of sophisticated vacation, science experiment, and public relations exercise, Edison observed the 1878 solar eclipse with his “tasimeter.” To calibrate the instrument, he observed and may have detected Arcturus [18]. However, the tasimeter was a very finicky and unstable instrument, probably usable only in the hands of its inventor, and the small size of the telescope and absence of thermal compensation in the instrument calls the detection into some question [64]. In any case, Edison had grander interests, and once the extensive publicity died down, there was no further effort.

The publicity for Edison’s feat did bring a rejoinder from Huggins, staking his claim for priority [35]. However, an attempt by Boys [6] to confirm the results of Huggins and Stone was not successful. Boys used a new detector, which he called a radiometer. It consisted of two matched thermocouples suspended in a current loop and inside a magnet. A small mirror attached to this delicate torsional pendulum (oscillation period of 10 s) allowed measuring its position with a light beam. Energy absorbed by one of the thermocouples modulated the current in the loop and caused a torque that moved the pendulum. Boys used a government grant of 50 lbs Sterling to build a telescope specifically around the requirements of this detector and those of measuring infrared radiation. He demonstrated that his detector was far more sensitive than those of Huggins and Stone. Nonetheless, he was unsuccessful at detecting any stars, but had to be satisfied by confirming Lord Rosse’s results on the heat radiation of the moon. He rather politely points out the issues (including Huggins’ use of a refractor) in describing the discrepancy [6]:

Dr. Huggins did not obtain any effect from the moon, but as he used an object glass which would absorb nearly all the moon’s heat or, at any rate, a large proportion, and as it had a smaller aperture and a greater focal length than the mirror which he lent me, so that the image of the moon was larger, and on this account also had a smaller heating power, it is not surprising that he did not obtain a large effect from the moon; but that he obtained none is, in the face of these measures, strong evidence that the deflections observed, and which he attributed to the stars, were spurious.

Detecting a star appears to have been a challenge taken up to demonstrate the performance of new types of infrared detector. The next attempt was by Nichols, who used a high performance version of Crookes’ Radiometer. This device survives as a novelty item: a set of vanes blackened on one side is placed inside a partially evacuated glass bulb, arranged so they can turn easily. When exposed to sunlight, the vanes in the novelty device spin rapidly due to the interaction of the heated surface with the gas in the bulb. Nichols used very light vanes of mica with a delicate suspension to improve over Boys’ device by about a factor of five in sensitivity [28]. Small signals were observed from Arcturus and Vega, in a ratio of about 2.1:1 in favor of Arcturus. Nonetheless, Hale [28] reports that:

In view of the smallness of the deflections, and the uncertainty which arises from rapid fluctuations in the atmosphere, Professor Nichols does not greatly rely upon the quantitative value of the results.

Thus, the nineteenth century saw the first efforts at detection of a star in the infrared (by Huggins, Stone, and Edison), the first government grant for infrared astronomy and the first infrared-optimized telescope (both by Boys), but ended with some finality that no astronomy had been accomplished.

3 A real beginning is ignored

But was that negative conclusion correct? Nichols [64] mulled over his measurements and decided he had achieved a detection after all. Indeed, if we reject the one measurement he describes as questionable, he finds a difference between Vega and Arcturus of 0.9 magnitudes, which can be compared with the value of 1.1 magnitudes measured 30 years later by Pettit and Nicholson [68] using equipment with similar spectral response. Nichols concludes:

The ratio greater than 2 to 1, of the total radiation of Arcturus to that of Vega (stars which by most observers are estimated to be of nearly equal photometric magnitude) indicates a proportionately more intense infra-red spectrum for the former than for the latter star. ... The photosphere of Arcturus may be at a lower temperature than that of Vega, but the star be of sufficiently greater angular diameter, as seen from the Earth, to equal Vega in light intensity and surpass it in total radiation.

We therefore give Nichols credit for the first stellar astronomical measurement in the infrared. However, his detector was clearly inadequate to open up a large field of investigation. Coblentz [8] solved this problem by dramatic improvements in a thermocouple-based detector. He realized that the best performance depended on achieving the smallest possible heat capacity along with good thermal isolation—the basic principles used to optimize thermal detectors in general. To achieve the first goal, he made single-junction detectors of only 0.3 to 0.4 mm size; placing the detector in a vacuum was an essential step toward the second goal. He also introduced a cell of water as a way to remove the infrared signal and deduce its strength through the change relative to the total signal with it out of the beam [13]). Thus, he provided a complete solution to the problems that had prevented much progress in the nineteenth century. Already by 1914 he had measured more than a hundred celestial objects, including a more accurate ratio of the outputs of Arcturus and Vega [8, 9]. He continued to improve detectors and filters (e.g., [10, 11, 12]), and determined a reasonable temperature scale for stars of different spectral types [14]. He clearly deserves the title of first infrared astronomer. Moreover, he set the scene for further advances that brought the field to a high state of development.

Pettit and Nicholson carried out an ambitious infrared astronomy program from about 1920, building on the technical foundation established by Coblentz. They began with a thorough optimization of the thermopile detectors [67]. They mounted a pair of these detectors on the Newtonian plate holder of the Mt. Wilson 100-in. Telescope, oriented so the observer could switch the stellar image between the thermocouple receivers by turning a guiding screw. The signals were conveyed by long wires to the basement of the telescope dome, where there was a sensitive galvanometer well-protected from any disturbances. In addition to a visible “real time” observer to monitor the readings and convey the results to the instrument operator at the telescope focus, the signals were permanently recorded on a photographic plate. As with Coblentz, the problem of isolating true infrared radiation was solved with a water cell; with 1 cm of water in the beam, virtually all radiation at wavelengths longer than ∼1.2 μm was blocked. The instrument had adequate sensitivity to reach hundreds of stars. It was sufficiently stable that the photometric data would be judged of reasonable quality by the standards of modern photometers (see the signals recorded in [67]). Comparison with modern commercial thermopiles (e.g. [81]) indicates that the devices employed by Pettit and Nicholson were an order of magnitude more sensitive. As we show below, their photometry was highly repeatable and quite accurate, although the photometric system had obvious shortcomings. For example, although the Harvard visible photometry indicated a difference between Sirius and Vega of 1.72 magnitudes, Pettit and Nicholson [68] measured a radiometric difference of only 1.37 magnitudes. The modern values are 1.46 magnitudes at V and 1.31 magnitudes in the near infrared.

Their masterwork addressed the then topical issue of stellar temperatures [68]. The paper is remarkable for the meticulous description of the derivation of accurate photometry, including corrections for effects such as the tarnishing of the telescope silver coatings. They also made observations with a specially prepared detector with a salt window to allow them to correct to total infrared flux from the cutoff near 3 μm imposed by the glass or quartz windows of the devices they normally used. Finally, their photographic recordings of the galvanometer output allowed precise reductions that could, for example, compensate for slow instrument baseline drifts, and could combine multiple photometric readings in a consistent way (they typically took eight for each observation).

From their measurements, Pettit and Nicholson [68] derived reasonably accurate stellar temperatures. A significant error source was the visible magnitudes, which were taken from the revised Harvard magnitudes [70]. These values represented a huge effort (more than a million individual measurements), obtained by direct visual observation without bandpass-defining filters. The effective wavelength was therefore uncertain. A comparison with modern V photometry indicates that a value significantly shorter than 0.555 μm is appropriate. Therefore, if we take the temperatures derived from the Heat-Index of Pettit and Nicholson and an assumed effective wavelength of 0.529 μm, we find that the agreement with modern values is of comparable accuracy as for other contemporary determinations (e.g., the tabulation they provide from Russell, Dugan, and Stewart). For cool stars, it is better.

In parallel with the work of Pettit and Nicholson, Abbott [1] used a low-dispersion prism to obtain infrared spectra of stars. His detector was based on the Crookes’ Radiometer; Nichols assisted him in the construction of early versions. To make the moving parts of very low mass and thereby boost the sensitivity, he substituted the wings of house flies for the mica plates Nichols had used as the radiation receivers. Although his published spectra show a general trend toward larger infrared emission for cooler stars, there is no clear pattern from star to star, even among the spectra he classifies as the best. It appears that the experimental difficulties were too great for his approach.

Pettit and Nicholson wrote a second very impressive paper, comparing the infrared and visible variability of cool stars [69]. Their two main works represented a high point of early infrared astronomy, with accurate and sensitive measurements, very careful analysis, and application to important astrophysical problems. Kuiper [42] used their work in computing what became the standard reference on stellar temperatures and bolometric corrections. Otherwise, neither paper seems to have received much attention until the 1950s. A few additional papers by others reported further radiometric measurements over the following decade, similar to the work of Coblentz and of Pettit and Nicholson [20, 80]. They had little impact. Most practicing astronomers considered infrared observations irrelevant.

4 Starting over again

Fifteen years after the work of Pettit and Nicholson, Whitford [89] obtained near-infrared measurements of a few stars to extend the interstellar extinction curve. The thermopiles had been replaced with a lead sulphide photoconductive cell, a spinoff of World War II. The galvanometer and photographic recording of its readings had also been replaced with vacuum-tube-based electronics. Bandpass definition remained a challenge; in this case, a dye-based filter isolated a band that merged the modern H and K bands. Whitford [90] later reported some additional measurements with the same apparatus. This work, although promising, was not extended beyond his specific interest in the form of the reddening curve. Kuiper et al. [43] also used a lead sulphide photocell in an infrared spectrometer and published stellar spectra, but thereafter turned the instrument to planetary studies.

Fellgett [21] also started an initiative in infrared astronomy. Despite the use of a lead sulphide photocell and vacuum-tube electronics, Fellgett remarks that the sensitivity of the photometer was only modestly improved over that of Pettit and Nicholson. The definition of photometric bands was also similar; a crude filter of mica blocked the visible but transmitted in the infrared to the cut-off wavelength of the detector. The accuracy of both systems is attested by Fellgett’s comparison of measurements of stars in common; it shows an average deviation between them of only 0.08 magnitudes. A surprising shortcoming of Fellgett’s photometer is the absence of any sky reference beam that would allow differential rather than total-power measurements. A differential approach had been found useful by Stone [85] and had been used in all the following photometers—including Boys, Nichols, Coblentz, and Pettit and Nicholson. Fellgett shows fluctuations in the signal from a star that he attributes to variations in atmospheric extinction, but he also claims that “it was often impossible to detect even considerable unsteadiness of the extinction by visual inspection of the sky.” It is likely that these fluctuations had a significant contribution from emission variations not compensated because his measurements did not include a simultaneous sensing of the sky.

However, unlike the compelling efforts to overcome all obstacles that infuses the papers of Pettit and Nicholson, Fellgett’s style suggests that his paper was intended to be a swan song. He rationalizes shortcomings in the equipment by saying they would be addressed only if a “first programme had shown whether a worthwhile infrared-red sensitivity would be realized.” On this issue, he proclaims, “there is little chance of detecting, in the infra-red, stars that are too cool to be visible.” He never published a sequel to his paper, and it is cited only for its historical significance.

Lunel [51] describes an extensive effort using an instrument similar in concept to Fellgett’s. However, an important improvement is that she utilized filters that cleanly isolated the H and K photometric bands, and also a band that roughly combined I and J. She measured 61 stars in this system and showed that the data correspond well with the observations of Pettit and Nicholson [68] and Fellgett [21]. Although the instrument and measurements are described in detail, her paper lacks the careful physical interpretation that is a strength of the work of Pettit and Nicholson [68]. Posterity has ignored it.

Moroz [57, 58, 59, 60] made the next start in infrared astronomy. His photometer was based on lead sulphide and germanium photocells, with spectral band definition not significantly advanced from the water cell of Coblentz. The most important innovation was his choice of non-stellar targets: the Crab Nebula, Galactic Center, and the Orion Nebula. The success of radio astronomy and the leadership of Russia in the theory of the newly discovered radio emissions had provided new classes of object for observation. However, Moroz’s efforts were largely overlooked in the rest of the world, part of a widespread indifference toward Russian observational astronomy as well as the indifference to infrared astronomy in general. Moroz had broad interests, and about the time his single-man infrared program began to be overrun by efforts in the USA, he found himself attracted into planetary exploration, where he spent the rest of his career.

Connes [16] developed high-spectral-resolution infrared spectrometers. A few stellar observations were obtained (e.g., [15]), which provided detailed lists of absorption features in cool stars [56]. However, the primary applications for these instruments were for planetary studies.

5 The beginning that took root

Modern infrared astronomy grew from a new start in the USA in the 1960s. The story is told in detail in Low et al. [50]. The first step was the lead-sulphide-based stellar photometry by Johnson [38], followed by expansion to the 10 μm atmospheric window by Wildey and Murray [91] and Low and Johnson [46]. Having seen so many false starts, it is important to consider what made this one different:
  1. 1.

    Despite what one might assume, it was NOT due to advances in sensitivity. Johnson’s [38] original set of measurements was in the same magnitude range, generally even a bit brighter, than the stars measured by Pettit and Nicholson [68]. The modest gains in photometer performance were nearly exactly cancelled by the use of smaller telescopes (as had previously been remarked by Fellgett [21]). Even Low’s [44] bolometer initially operated at similar detection levels to the previously available mid-infrared photoconductors (compare the results of Wildey and Murray [91] with those of Low and Johnson [45, 46]).

  2. 2.

    There were a number of important technical advances, nonetheless: (a) electronic integration allowed long accumulation of signals with full realization of the intrinsic system signal-to-noise potential; and (b) two decades of improvements in filter technology had led to high-performance infrared interference filters [52]), critical to isolate well-defined photometric bands matched to the atmospheric windows. Although Johnson originally used total-power chopping, the huge sky emission at 10 μm required differential measurements ([45, 91]), leading to rediscovery of their value originally established a century earlier ([79, 85]). Improved modulation capabilities were the first critical innovation introduced as part of the growth of the field; they let the improving detectors actually function at their full performance levels.

There was a third and far more important element. The brilliant success of radio astronomy demonstrated the scientific potential for astronomy in new spectral regions. The new start in infrared astronomy was led almost entirely by experimental physicists, not by astronomers [50]. They found infrared astronomy to be an open playing field where a bit of ingenuity would yield many exciting discoveries. As with Moroz, they were strongly influenced in their selection of targets by discoveries in the radio regime.
The development of infrared astronomy primarily by physicists, even in the face of indifference and antipathy in traditional astronomy departments, also had a precedent in radio astronomy. Jarrell [36] summarizes the backgrounds of 95 early radio astronomers, finding there were 61 physicists, 21 engineers, two mathematicians, and only 11 astronomers. Publication in the new field was often in physics or neutral journals (such as Nature). Citations of these papers by proper astronomers was rare; Jarrell [36] summarizes a study of citation rates: “To be sure, several important optical and theoretical astronomers....showed interest in the early radio work. But, for most of their colleagues, radio data were simply irrelevant to their research. In fact, radio results were as likely to be of as much value to atmospheric scientists as to astronomers.” The situation led to the following editorial:

Radio astronomy is a rapidly developing branch of astronomy; and any schism between this newer branch and the older branch of optical astronomy is to be greatly regretted. For this reason, the Astronomical Journal, already established as one of the principal astronomical journals, would like to see that papers on radio astronomy, commensurate with its growing importance, will be represented in the Astronomical Journal.”—Brouwer and Smith [7]

A parallel and similarly worded editorial also appeared in the Astrophysical Journal.

X-ray astronomy also grew out of physics and other efforts not linked to traditional astronomy departments ([87]; Giacconi 2008, private communication). There was evidently a general resistance to initiatives into new spectral regimes on the part of most twentieth century optical astronomers. Thus, we find a fourth key element to the successful growth of infrared astronomy from the new start in the early 1960s. By chance, there were enough centers of activity for a community to be established—to be sure a community split by intense rivalries, but still where one’s accomplishments would be appreciated, sometimes envied, and rapidly incorporated into other scientific and technical advances.

In the decade from the early 1960s to early 1970s, virtually the entire foundation for infrared astronomy was put in place, such as:
  1. 1.

    Establishment of the photometric system, its calibration, and application of it to large numbers of stars [39]

  2. 2.

    Completion of an all-sky survey that successfully disproved Fellgett’s discouraging prediction that “there is little chance of detecting, in the infra-red, stars that are too cool to be visible.” [62]

  3. 3.

    Detection of the internal energy of Jupiter and the other giant planets [3]

  4. 4.

    Identification of oxygen-rich and carbon-rich interstellar dust, the former in a range of environments [53, 84, 92]

  5. 5.

    Discovery of the prototypical dust-embedded very young star [4]

  6. 6.

    Finding the huge and ubiquitous far-infrared outputs of regions where young and massive stars are forming [33, 48]

  7. 7.

    Discovery of the center of the Milky Way, both in stellar emission [5] and in thermal re-emission of the stellar energy by interstellar dust [31]

  8. 8.

    Detection of the strong infrared outputs of star-forming galaxies[41] and discovery of Luminous Infrared Galaxies and Ultraluminous Infrared Galaxies (LIRGs and ULIRGs) [75]

  9. 9.

    Demonstration of the infrared excesses associated with active galactic nuclei [47, 65]

In parallel, the prime approaches toward infrared observations were established:
  1. 1.

    Precepts were laid out for optimizing ground-based telescopes for good infrared performance [49]

  2. 2.

    Airborne infrared astronomy got under way [17]

  3. 3.

    The advantages of operating cold telescopes in the space environment were demonstrated (e.g. [33, 72])


6 Building on the foundation

The sky had been super-saturated with potential discoveries in the infrared. That they had not been made earlier, when some of the technical means to do so were already in place, runs contrary to our usual assumption that discoveries follow quickly after we have the capability to make them [30]. Human imagination had, for once, come up short. However, the explosion of discoveries in the 1960s and early 1970s attracted the attention of the international astronomy community. The instrumentation required to join in was quite modest. Therefore, infrared programs rapidly sprang up in the UK [37], Europe, Australia [86], and South Africa. The discoveries also attracted the interest of advisory bodies (e.g., the National Academy “Greenstein” report) and government funding agencies. Soon, there were plans for substantial upgrades in all three main observing approaches. These upgrades were fuelled by incredible technical progress that occurred, ironically, after the field had taken root rather than before.

6.1 Groundbased telescopes

A key step in the USA was a meeting in July, 1975, of an ad hoc advisory group convened by Nancy Boggess (who oversaw the interests of infrared astronomy for NASA) at the Snowmass Colorado ski resort. This group urged NASA to proceed with a large, national infrared telescope. The resulting Infrared Telescope Facility (IRTF) was completed in 1979. In the interim, Gehrz and Hackwell [24] had persuaded the Wyoming legislature and NSF to fund a 2.3-m infrared optimized telescope. Both efforts were trumped by the UK, which completed the UK Infrared Telescope (UKIRT) also in 1979. Thus, by 1980 the astronomical community had access to a number of infrared telescopes at excellent sites (low water vapor).

In 1975, InSb photodiodes were introduced with performance far better than lead sulphide cells [29]. Otherwise, the detectors, optics, and instrumentation evolved slowly from those used in the 1960s. With telescopes virtually as large as possible at the time, it appeared that the rapid growth of the field might level off. Instead, it accelerated due to the introduction of infrared detector arrays, plus the success of the Infrared Astronomical Satellite (IRAS).

Rudimentary arrays became available in the early 1980s. The challenges in the mid-infrared, with the immense signal from the thermal background, proved intractable for sensitive operation, although a few images of bright objects were produced [2]. However, in the near infrared even the very early arrays worked reasonably well. The first example was a 32 × 32 pixel InSb array provided through Alan Hoffman of Santa Barbara Research Center to the University of Rochester, where Hoffman had been an Assistant Professor before joining SBRC [22]. A similar array but with HgCdTe detectors and from Rockwell Science Center was deployed at the University of Arizona shortly thereafter [78]. In both cases, the array readout circuits were charge-coupled-devices (CCDs). CCDs are fundamentally incompatible with high-performance infrared detectors; the detectors require operating temperatures so low that the CCDs must operate in surface channel mode, and hence the read noises of these devices were high (>1,000 electrons). Nonetheless, the performance of individual pixels was comparable to that of the best single detectors. The readout limitations were soon removed by development of devices that had a readout amplifier hidden behind each detector and hence were not subject to the limitations of charge transfer. The first widely applied array of this type, from SBRC, used a readout of 58 × 62 format fitted with a variety of detector types.

All of these initial detector arrays were developed for other (usually military) customers. They were followed by arrays financed by the NSF and NASA and optimized for astronomical applications–slow readout, low noise, large format, and high quantum efficiency. These developments were fostered by the enthusiasm of the array manufacturers. Most of their other customers operated on a principle of compartmentalization of information, so it was not clear what use, if any, was intended for the arrays. Astronomy was wide open; the intended uses were public and interesting, and the customers highly interactive on the design details. A number of astronomers even migrated into array manufacturing. As a result of this partnership, the funding from NSF and NASA was used very effectively to yield the current state of the art devices [77]. A single pixel of one of these megapixel arrays is two to three orders of magnitude more sensitive than the single detectors used 30 years ago.

Soon after the first astronomy-optimized arrays became available, the first 6–10 m telescopes were completed. With the application of adaptive optics, increasingly advancing to a general-purpose capability, these telescopes routinely provide diffraction-limited infrared images. Compared with capabilities 40 years ago, the net gain across the 1–25 μm spectral range is factors of 1,000 to more than 10,000 in sensitivity, a gain in multiplex capability of up to four million, and enhancement of the angular resolution (compared with typical aperture photometers) of two orders of magnitude.

7 Airborne astronomy

Frank Low’s initiation of airborne infrared astronomy was a natural step from: (1) his previous measurement of the mm-wave output of the sun from a Navy Douglas A3-B bomber; and (2) his position at the University of Arizona Lunar and Planetary Laboratory, where director Gerard Kuiper and his group made use of the CV-990 Galileo I Airborne Observatory for planetary observations. Low and Carl Gillespie fitted a 12-in. telescope into the emergency exit of a Lear Jet at NASA Ames Research Center. On their first flight series, the telescope was not well stabilized and there were only two accessible sources bright enough for guiding—Arcturus and Jupiter. The observations revealed the internal energy of the planet–airborne astronomy started with a fanfare! The Lear Jet Observatory continued to fly until 1984 [17]).

However, a much larger telescope was desired. Plans were developed for a 36-in. (0.9 m). The C-141 cargo airplane to carry it was delivered in early 1972 and research flights started 2 years later. In May, 1975, the facility was named the Kuiper Airborne Observatory (KAO). Observing with the Lear Jet had been a barebones adventure, wearing an oxygen mask and squeezed between the back of the telescope and the opposite wall of the fuselage. It conjured up self-images of having the Right Stuff. The KAO ushered in the shirt-sleeve era of airborne infrared astronomy, an environment more conducive to demanding observing. It flew about 70 flights per year for more than 20 years, both for infrared astronomy and for other applications where a highly mobile observatory was required (e.g., the discovery of the rings of Uranus [17, 19]).

Balloons have been intermittently popular to lift infrared telescopes above the atmosphere. Early near-infrared spectra were obtained with the Stratoscope II telescope [93]. We have already mentioned the discovery of the far-infrared emission of the Galactic Center by Hoffmann and Frederick [31]; this group made a number of additional flights to map the region in detail (e.g., [32]). Shortly thereafter, a 1-m balloon was deployed (e.g. [94]). Subsequent efforts have led up to the BLAST, a 2-m telescope with a high performance bolometer imaging array [66]. However, in general the far-infrared community has had a preference for airplane-borne telescopes, in part because such observatories have been more successful at establishing a long-term infrastructure than have balloons.

The ultimate airborne observatory would use the largest possible airplane. NASA approved a joint project with Germany to build just that in 1996. Named the Stratospheric Observatory for Infrared Astronomy (SOFIA), the new observatory would utilize a 2.5-m telescope in a Boeing 747. In part to avoid splitting the infrared astronomy community, SOFIA and Spitzer were sold as a package, and in an era of very tight funding. The plans for SOFIA also further enhanced the shirt-sleeve environment of the KAO. Perhaps lulled by the very rapid deployment of the KAO, astronomers agreed to decommission that observatory to free funds for the development of SOFIA.

The modifications to the 747 airframe proved more complex to implement than expected. In addition, the desire to certify the entire observatory to passenger transport standards added substantial complexity to the instruments and other aspects of the project. The schedule and costs grew until 2006, when after a period of review, a brief halt, and threatened cancellation, work eventually was resumed. Science observations are now expected to begin with SOFIA in 2009–2010. At that time, the observatory will offer the far-infrared community the major advantage of a readily accessible telescope, with the ability to update instrumentation regularly to keep the observatory at the state of the art and to introduce and demonstrate new technologies [26]. SOFIA is planned for a 20 year service life.

8 Space infrared astronomy

The advantages of operating a cold infrared telescope above the atmosphere were realized in early sounding-rocket experiments at Cornell under Martin Harwit and Jim Houck [50], and through a variety of missions funded by the military and described by Price [72]. They included HISTAR, operated by Price from the Air Force Geophysical Laboratory (AFGL) and the first publicly available mid-infrared survey (a satellite-borne effort, the Celestial Mapping Program, had been released previously but in an irregular fashion in the form of a photographic reproduction of an all-sky infrared source locator).

Besides a large groundbased telescope, the Snow Mass meeting recommended an ambitious all-sky survey to cover the mid and far infrared. In parallel, the Netherlands had studied the feasibility of a far-infrared satellite telescope. NASA issued an Announcement of Opportunity that attracted a number of proposals. They were combined with a down-selection of members and an agreement was negotiated in which the Dutch would provide the spacecraft, the United States the telescope, data system, and dewar, and the United Kingdom the ground station. The resulting IRAS mission is described by Neugebauer et al. [63]. It was one of the most successful space astronomy missions ever conducted and remains a fundamental resource more than 20 years after it was completed.

The success of IRAS led to efforts in both Europe and the United States to build a follow-up mission in the form of a pointed cryogenic telescope that could concentrate on individual sources. A proposal had already been prepared in Europe in 1979, and was accepted as a cornerstone European Space Agency (ESA) mission as soon as the success of IRAS became apparent. Construction started in 1988, with a launch in 1995 and a mission that lasted for more than 2 years. ISO (the Infrared Space Observatory) was the first infrared astronomy mission to use true detector arrays. Its suite of instruments included many capabilities, such as high resolution spectrometers, to enable detailed astrophysical studies. An overview of the mission is given by Kessler [40], and a review of the extragalactic studies by Genzel and Cesarsky [27]).

The US effort ran into repeated political setbacks [76] and was finally launched in 2003, and renamed Spitzer. It had been descoped into a much simpler mission than originally envisioned, based on a pioneering concept of developing the design around a small number of defining science programs. It also introduced a new cooling concept that placed heavy reliance on radiative cooling with only a supplemental capacity to reach the final telescope operating temperature and cool the detectors; the mission cryogenic lifetime has been more than 5 years. Fortunately, the development of very high performance infrared detector arrays allowed a mission with superb performance and that, like ISO, has yielded many science breakthroughs. The mission design is described by Gehrz et al. [25], and early reviews of the science accomplished are provided by Werner et al. [88] and Soifer et al. [83].

The Akari Mission was developed by the Japan Aerospace Exploration Agency (JAXA) and completed a double-pass all-sky survey in a number of bands from 1.7 to 180 μm, over its 18 month cryogenic mission (see PASJ [73]). It also had a program of pointed observations both with imagers and with additional instrument capabilities, primarily spectroscopy. The success of this mission underlies the proposal from Japan for a much more ambitious far-infrared cold telescope, the Space Infrared Telescope for Cosmology and Astrophysics (SPICA) [55, 61]. A more immediate advance in far-infrared astronomy will be provided by Herschel, due to be launched by the European Space Agency in mid-2009. It is a large-aperture (3.5 m) radiatively cooled (to about 100 K) far-infrared mission with three instruments providing imaging and a range of spectroscopic capabilities from 60 to 670 μm [71]. Herschel will be the first long-duration far-infrared observatory with a telescope aperture greater than a meter! Also planned for launch in 2009, the Wide-field Infrared Survey Explorer (WISE) will map the sky at 3.3, 4.7, 12, and 23 μm [54].

However, the realization of a dream for infrared astronomers will be the James Webb Space Telescope (JWST), of 6.6 m aperture and with four powerful focal plane instruments [23]. It will cover the 0.7 to 28 μm wavelength range and due to the radiative cooling of its primary to about 40 K will be natural-background limited over most of this wavelength region. Upon the launch of this mission in 2013, the astronomical capability in the infrared—number of pixels/(detection limit squared)—will have increased by a factor of 1020 over 50 years, a doubling time of 9 months sustained over this entire time span.



Preparation of this article was partially supported by contract 1255094 from Caltech/JPL and by contract NAG5-12318 from NASA/Goddard, both to the University of Arizona.


  1. 1.
    Abbott, C.G.: Energy spectra of the stars. ApJ 69, 293 (1929)CrossRefADSGoogle Scholar
  2. 2.
    Arens, J.F., Lamb, G.M., Peck, M.C., Moseley, H., Hoffmann, W.F., Tresch-Fienberg, R., Fazio, G.G.: High spatial resolution observations of NGC 7027 with a 10 micron array camera. ApJ 279, 685 (1984)CrossRefADSGoogle Scholar
  3. 3.
    Aumann, H.H., Gillespie, C.M., Low, F.J.: The internal powers and effective temperatures of Jupiter and Saturn. ApJL 157, L69 (1969)CrossRefADSGoogle Scholar
  4. 4.
    Becklin, E.E., Neugebauer, G.: Observations of an infrared star in the Orion nebula. ApJ 147, 799 (1967)CrossRefADSGoogle Scholar
  5. 5.
    Becklin, E.E., Neugebauer, G.: Infrared observations of the galactic center. ApJ 151, 145 (1968)CrossRefADSGoogle Scholar
  6. 6.
    Boys, C.V.: On the heat of the moon and stars. Proc. R. Soc. 47, 480 (1890)CrossRefGoogle Scholar
  7. 7.
    Brouwer, D., Smith, H.J.: Publication of papers in radio astronomy. AJ 64, 36 (1959)CrossRefGoogle Scholar
  8. 8.
    Coblentz, W.W.: Note on the radiation from stars. Publ. Astron. Soc. Pac. 26, 169 (1914)CrossRefADSGoogle Scholar
  9. 9.
    Coblentz, W.W.: A Comparison of Stellar Radiometers and Radiometric Measurements on 110 Stars. Government Printing Office, Washington (1915)Google Scholar
  10. 10.
    Coblentz, W.W.: Report on instruments and methods of radiometry. J. Opt. Soc. Am. 5, 259 (1921)CrossRefADSGoogle Scholar
  11. 11.
    Coblentz, W.W.: The measurement of solar, sky, nocturnal, and stellar radiation. J. Opt. Soc. Am. 5, 269 (1921)CrossRefADSGoogle Scholar
  12. 12.
    Coblentz, W.W.: A portable vacuum thermopile. J. Opt. Soc. Am. 5, 356 (1921)CrossRefADSGoogle Scholar
  13. 13.
    Coblentz, W.W.: New measurements of stellar radiation. ApJ 55, 20 (1922)CrossRefADSGoogle Scholar
  14. 14.
    Coblentz, W.W.: The effective temperature of 16 stars as estimated from the energy distribution in the complete spectrum. Proc. Natl. Acad. U. S. A. 8, 49 (1922)CrossRefADSGoogle Scholar
  15. 15.
    Connes, P., Connes, J., Bouigue, R., Querci, M., Chauville, J., Querci, F.: Sur les spectres à étoiles rouges M et C entre 4000 et 9000 cm − 1 (2,5 et 1,1 μ). I. Description génélrale. ANAP 31, 485 (1968)ADSGoogle Scholar
  16. 16.
    Connes, P.: Early history of Fourier transform spectroscopy. IR Phys. 24, 69 (1984)Google Scholar
  17. 17.
    Dolci, W.W.: Milestones in Airborne Astronomy: From the 1920’s to the Present. Am. Inst. Aeronautics and Astronautics, Reston, Virginia (1997)Google Scholar
  18. 18.
    Eddy, J.A.: Thomas A. Edison and infrared-red astronomy. J. Hist. Astron. 3, 165 (1972)ADSGoogle Scholar
  19. 19.
    Elliot, J.L.: The rings of Uranus. Nature 267, 328 (1977)CrossRefADSGoogle Scholar
  20. 20.
    Emberson, R.M.: Radiometric magnitudes of some of the brightest stars. ApJ 94, 427 (1941)CrossRefADSGoogle Scholar
  21. 21.
    Fellgett, P.B.: An exploration of infrared stellar magnitudes using the photo-conductivity of lead sulphide. MNRAS 111, 537 (1951)ADSGoogle Scholar
  22. 22.
    Forrest, W.J., Moneti, A., Woodward, C.E., Pipher, J.L., Hoffman, A.: The new near infrared rray camera at the University of Rochester. Publ. Astron. Soc. Pac. 97, 183 (1985)CrossRefADSGoogle Scholar
  23. 23.
    Gardner, J.P. et al.: The James Webb space telescope. Sp. Sci. Rev. 123, 485 (2006)CrossRefADSGoogle Scholar
  24. 24.
    Gehrz, R.D., Hackwell, J.A.: Exploring the infrared universe from Wyoming. Sky Telesc. 55, 466 (1978)ADSGoogle Scholar
  25. 25.
    Gehrz, R.D., et al.: The NASA Spitzer space telescope. Rev. Sci. Inst. 78, 011032 (2007)Google Scholar
  26. 26.
    Gehrz, R.D., Becklin, E.E.: The stratospheric observatory for infrared astronomy (SOFIA). Proc. SPIE 7012, 61 (2008)ADSGoogle Scholar
  27. 27.
    Genzel, R., Cesarsky, C.J.: Extragalactic results from the infrared space observatory. ARAA 38, 761 (2000)CrossRefADSGoogle Scholar
  28. 28.
    Hale, G.E.: Heat radiation of the stars. ApJ 9, 360 (1899)CrossRefADSGoogle Scholar
  29. 29.
    Hall, D.N.B., Aikens, R.S., Joyce, R., McCurnin, T.W.: Johnson noise limited operation of photovoltaic InSb detectors. Appl. Opt. 14, 450 (1975)CrossRefADSGoogle Scholar
  30. 30.
    Harwit, M.W.: Cosmic Discovery. Basic Books, New York (1981)Google Scholar
  31. 31.
    Hoffmann, W.F., Frederick, C.L.: Far-infrared observation of the galactic-center region at 100 microns. ApJL 155, L9 (1969)CrossRefADSGoogle Scholar
  32. 32.
    Hoffmann, W.F., Frederick, C.L., Emery, R.J.: Far-infrared observation of the galactic-center region at 100 microns. ApJL 170, L89 (1971)CrossRefADSGoogle Scholar
  33. 33.
    Houck, J.R., Soifer, B.T., Pipher, J.L., Harwit, M.: Rocket-infrared four-color photometry of the galaxy’s central regions. ApJL 169, 31 (1971)CrossRefADSGoogle Scholar
  34. 34.
    Huggins, W.: Note on the heat of the stars. Proc. Royal Soc. 17, 309 (1869)CrossRefGoogle Scholar
  35. 35.
    Huggins, W.: Heat of the stars. Astron. Regist. 16, 309 (1878)ADSGoogle Scholar
  36. 36.
    Jarrell, R.: Radio astronomy, whatever that may be. In: Orchiston, W. (ed.) The Marginalization of Early Radio Astronomy. Astrophysics and Space Science Library, vol. 334. Springer, Dordrecht (2005)Google Scholar
  37. 37.
    Jennings, R.E.: History of British infrared astronomy since the second world war. QJRAS 27, 454 (1986)ADSGoogle Scholar
  38. 38.
    Johnson, H.L.: Infrared stellar photometry. ApJ 135, 69 (1962)CrossRefADSGoogle Scholar
  39. 39.
    Johnson, H.L.: Astronomical measurements in the infrared. ARAA 4, 193 (1966)CrossRefADSGoogle Scholar
  40. 40.
    Kessler, M.: The infrared space observatory (ISO) mission. Adv. Space Res. 30, 1957 (2002)CrossRefADSGoogle Scholar
  41. 41.
    Kleinmann, D.E., Low, F.J.: Observations of infrared galaxies. ApJL 159, L165 (1970)CrossRefADSGoogle Scholar
  42. 42.
    Kuiper, G.P.: The magnitude of the sun, the stellar temperature scale, and bolometric corrections. ApJ 88, 429 (1938)CrossRefADSGoogle Scholar
  43. 43.
    Kuiper, G.P., Wilson, W., Cashman, R.J.: An infrared stellar spectrometer. ApJ 106, 243 (1947)CrossRefADSGoogle Scholar
  44. 44.
    Low, F.J.: Low-temperature germanium bolometer. J. Opt. Soc. Am. 51, 1300 (1961)CrossRefADSGoogle Scholar
  45. 45.
    Low, F.J., Johnson, H.L.: Stellar photometry at 10 microns. ApJ 139, 1130 (1964)CrossRefADSGoogle Scholar
  46. 46.
    Low, F.J., Johnson, H.L.: The spectrum of 3C273. ApJL 141, L336 (1965)CrossRefGoogle Scholar
  47. 47.
    Low, F.J., Kleinmann, D.E.: Infrared observations of seyfert galaxies, quasistellar sources, and planetary nebulae. AJ 73, 868 (1968)CrossRefADSGoogle Scholar
  48. 48.
    Low, F.J., Aumann, H.H.: Observations of galactic and extragalactic sources between 50 and 300 microns. ApJL 162, 79 (1970)CrossRefADSGoogle Scholar
  49. 49.
    Low, F.J., Rieke, G.H.: The instrumentation and techniques of infrared photometry. In: Carleton, N. (ed.) Methods of Experimental Physics, vol. 12, part A. Academic Press, New York and London (1974)Google Scholar
  50. 50.
    Low, F.J., Rieke, G.H., Gehrz, R.D.: The beginning of modern infrared astronomy. ARAA 45, 43 (2007)CrossRefADSGoogle Scholar
  51. 51.
    Lunel, M.: Recherches de photométrie stellaire dans l’infra-rouge au moyen d’une cellule au sulfure de plomb. Ann. D’Astrophysique. 23, 1 (1960)ADSGoogle Scholar
  52. 52.
    Macleod, H.A.: Thin-Film Optical Filters, 3rd edn. Bristol, Philadelphia (2001)Google Scholar
  53. 53.
    Maas, R.W., Ney, E.P., Woolf, N.J.: The 10-micron emission peak of comet Bennett 1969I. ApJL 160, L101 (1970)CrossRefADSGoogle Scholar
  54. 54.
    Mainzer, A.K., et al.: Preliminary design of the wide-field infrared survey explorer (WISE). Proc. SPIE 5899, 262 (2005)ADSGoogle Scholar
  55. 55.
    Matsumoto, T.: Large-aperture cooled telescope (SPICA) for mid- and far-infrared astronomy. Proc. SPIE 5487, 1501 (2004)CrossRefADSGoogle Scholar
  56. 56.
    Montgomery, E.F., Connes, P., Connes, J., Edmonds, F.N.: The infrared spectrum of Arcturus. ApJ 19, 1 (1969)CrossRefADSGoogle Scholar
  57. 57.
    Moroz, V.I.: The radiation flux from the Crab Nebula at 2 microns and some conclusions on the spectrum and magnetic field. Astron. Z. 37, 265 (1960)ADSGoogle Scholar
  58. 58.
    Moroz, V.I.: An attempt to observe the infrared radiation of the galactic nucleus. Astron. Z. 38, 487 (1961)ADSGoogle Scholar
  59. 59.
    Moroz, V.I.: Radiation emission from the Orion Nebula in the 0.85–1.7 micron wavelength region. Astron. Z. 40, 788 (1963)ADSGoogle Scholar
  60. 60.
    Moroz, V.I.: Infrared observations of the Crab Nebula. Astron. Ah. 40, 982 (1963)ADSGoogle Scholar
  61. 61.
    Nakagawa, T., Murakami, H.: Mid- and far-infrared astronomy mission SPICA. Adv. Space Res. 40, 679 (2007)CrossRefADSGoogle Scholar
  62. 62.
    Neugebauer, G., Leighton, R.B.: Two-micron sky survey. A preliminary catalogue. NASA Spec. Pub. 3047 (1969)Google Scholar
  63. 63.
    Neugebauer, G., et al.: The infrared astronomical satellite (IRAS) mission. ApJL 278, L1 (1984)CrossRefADSGoogle Scholar
  64. 64.
    Nichols, E.F.: On the heat radiation of Arcturus, Vega, Jupiter, and Saturn. ApJ 13, 101 (1901)CrossRefADSGoogle Scholar
  65. 65.
    Oke, J.B., Neugebauer, G., Becklin, E.E.: Absolute spectral energy distribution of quasi- stellar objects from 0.3 to 2.2 microns. ApJ 159, 341 (1970)CrossRefADSGoogle Scholar
  66. 66.
    Pascale, E. et al.: The Balloon-borne large aperture submillimeter telescope: BLAST. ApJ 681, 400 (2008)CrossRefADSGoogle Scholar
  67. 67.
    Pettit, E., Nicholson, S.B.: Application of vacuum thermocouples to problems in astrophysics. ApJ 56, 295 (1922)CrossRefADSGoogle Scholar
  68. 68.
    Pettit, E., Nicholson, S.B.: Stellar radiation measurements. ApJ 68, 279 (1928)CrossRefADSGoogle Scholar
  69. 69.
    Pettit, E., Nicholson, S.B.: Measurements of the radiation from variable stars. ApJ 78, 320 (1933)CrossRefADSGoogle Scholar
  70. 70.
    Pickering, E.C.: Revised Harvard photometry. Harvard Annals 50, 1 (1908)ADSGoogle Scholar
  71. 71.
    Pilbratt, G.: Herschel mission: status and observing opportunities. SPIE 5487, 401 (2004)CrossRefADSGoogle Scholar
  72. 72.
    Price, S.D.: The infrared sky: a survey of surveys. Publ. Astron. Soc. Pac. 100, 171 (1988)CrossRefADSGoogle Scholar
  73. 73.
    Publ. Astron. Soc. Jpn. (Akari Special Issue) 59, SP2 (2007)Google Scholar
  74. 74.
    Ranyard, A.C.: Edison’s tasimeter. Astron. Regist. 16, 309 (1878)ADSGoogle Scholar
  75. 75.
    Rieke, G.H., Low, F.J.: Infrared photometry of extragalactic sources. ApJL 176, L95 (1972)CrossRefADSGoogle Scholar
  76. 76.
    Rieke, G.H.: The Last of the Great Observatories. University of Arizona, Tucson (2006)Google Scholar
  77. 77.
    Rieke, G.H.: Infrared detector arrays for astronomy. ARAA 45, 77 (2007)CrossRefADSGoogle Scholar
  78. 78.
    Rieke, M.J., Rieke, G.H., Montgomery, E.F.: Rockwell HgCdTe arrays as imagers. In: Wynn-Williams, C.G., Becklin, E.E. (eds.) Infrared Astronomy with Arrays, p. 213. University of Hawaii Institute for Astronomy (1987)Google Scholar
  79. 79.
    Rosse, E.O: On the radiation of heat from the moon, the law of its absorption by our atmosphere, and its variation in amount with her phases. Proc. R. Soc. 21, 241 (1873)CrossRefGoogle Scholar
  80. 80.
    Rust, C.F.: Spectral types and radiometric observations of stars with large infrared index. ApJ 88, 525 (1938)CrossRefADSGoogle Scholar
  81. 81.
    Schilz, J.: Thermoelectric Infrared Sensors (Thermopiles) for Remote Temperature Measurements. Perkin Elmer Optoelectronics GmbH, Wiesbaden, Germany (2000)Google Scholar
  82. 82.
    Sinton, W.M.: Through the infrared with logbook and lantern slides: a history of infrared astronomy from 1868 to 1960. Publ. Astron. Soc. Pac. 98, 246 (1986)CrossRefADSGoogle Scholar
  83. 83.
    Soifer, B.T., Helou, G., Werner, M.W.: The Spitzer view of the extragalactic universe. ARAA 46, 201 (2008)CrossRefADSGoogle Scholar
  84. 84.
    Stein, W.A., Gillett, F.C.: Spectral distribution of infrared radiation from the trapezium region of the Orion Nebula. ApJL 155, L197 (1969)CrossRefADSGoogle Scholar
  85. 85.
    Stone, E.F.: Approximate determinations of the heating-powers of Arcturus and Alpha Lyrae. Proc. R. Soc. 18, 159 (1870)CrossRefGoogle Scholar
  86. 86.
    Storey, J.W.V.: Infrared astronomy: in the heat of the night. Publ. Astron. Soc. Aust 17, 270 (2000)ADSGoogle Scholar
  87. 87.
    Tucker, W., Giacconi, R.: The X-ray Universe. Harvard University Press, Cambridge (1985)Google Scholar
  88. 88.
    Werner, M.W., Fazio, G.G., Rieke, G.H., Roellig, T.L., Watson, D.M.: First fruits of the Spitzer space telescope: galactic and solar system studies. ARAA 44, 269 (2006)CrossRefADSGoogle Scholar
  89. 89.
    Whitford, A.E.: An extension of the interstellar absorption-curve. ApJ 107, 102 (1948)CrossRefADSGoogle Scholar
  90. 90.
    Whitford, A.E.: The law of interstellar reddening. AJ 63, 201 (1958)CrossRefADSGoogle Scholar
  91. 91.
    Wildey, R.L., Murray, B.C.: Stellar photometry from B8 to M7 in the 8–14 micron window. AJ 68, 300 (1963)CrossRefADSGoogle Scholar
  92. 92.
    Woolf, N.J., Ney, E.P.: Circumstellar infrared emission from cool stars. ApJL 155, L181 (1969)CrossRefADSGoogle Scholar
  93. 93.
    Woolf, N.J., Schwarzschild, M., Rose, W.K.: Infrared spectra of red-giant stars. ApJ 140, 833 (1964)CrossRefADSGoogle Scholar
  94. 94.
    Wright, E.L., Fazio, G.G., Low, F.J.: A high-resolution far-infrared survey of the W31 region. ApJ 217, 724 (1977)CrossRefADSGoogle Scholar

Copyright information

© Springer Science+Business Media B.V. 2009

Authors and Affiliations

  1. 1.Steward ObservatoryThe University of ArizonaTucsonUSA

Personalised recommendations