The fable of the blind men and the elephant has much to teach us about scientific and philosophical endeavors of all kinds. Yet few undertakings approximate the blind men’s predicament more closely than the search for extrasolar planets. In this variation on the tale, the blind men stand in for astronomers, the elephant represents the immense population of exoplanets spinning all around us, and the features of its anatomy that the blind men scrutinize are the data resulting from the various observational techniques used by planet hunters.

To date, 5 different methods have succeeded in detecting exoplanets: direct imaging, pulsation timing, transit observations, gravitational microlensing, and radial velocity measurements. Although each method has revealed both single- and multi-planet systems, each one also has powerful biases that limit its effectiveness to a subset (sometimes quite small) of the anticipated multitude of extrasolar planets.

Direct imaging is the old-fashioned way of doing astronomy. Unlike other detection methods, it relies on capturing visible or infrared light originating in the target exoplanet itself, rather than on measuring the characteristics and behavior of its host star. Because the contrast in luminosity between star and planet is so extreme – a common metaphor likens the planet and its parent star to a firefly caught in the glare of a searchlight – direct imaging suffers from substantial limitations. These shortcomings constrain its effectiveness to gas giants orbiting at wide separations around very young and relatively nearby stars (typically of spectral type A). The most notable successes of direct imaging have been the recent detection of massive planets around Fomalhaut, HR 8799, and Beta Pictoris (Kalas et al. 2008, Marois et al. 2008, Lagrange et al. 2010), all of which are A-type stars located within 40 parsecs (130 light years).

The HR 8799 planetary system. HR 8799 is a young A5 star located 39.4 parsecs (128 light years) away. Its 3 planets were directly imaged in 2008 by a team led by Christian Marois. Designated b, c, and d, they orbit at projected separations of 68, 38, and 24 AU, respectively, with masses estimated between 7 and 10 times Jupiter (Marois et al. 2008). Image credit: C. Marois et al./NRC Canada/Keck Observatory.

Pulsation timing measures periodic variations in the signal emitted by pulsating neutron stars (pulsars), pulsating B-type subdwarfs, and pulsating white dwarfs, all of which are rare in our galactic neighborhood. To date, this method has revealed gas giant and Super Earth-type planets in 2 pulsar systems, the closer of which is 300 parsecs (almost 1000 light years) distant; massive gas giants around 3 subdwarfs at similar distances; and one potential gas giant orbiting a white dwarf (GD 66) about 51 parsecs (166 light years) away (Mullally et al. 2009). Although pulsation timing was responsible for the very first exoplanet discovery in 1990 (Wolszczan & Frail 1992), its limited applicability prevents widespread use in search programs.

Transit observations record regular dips in the brightness of candidate stars caused by orbiting planets that are fortuitously aligned so that we observe them in transit across the face of their parent stars. So far this approach has helped to characterize more than 80 exoplanets (Charbonneau et al. 2007, Torres et al. 2008, Extrasolar Planets Encyclopaedia [EPE]). However, transit searches are strongly biased in favor of objects with large diameters traveling on very short-period orbits around bright stars. As a result, almost 90% of the known transiting planets are Hot Jupiters – gas giants with periods shorter than 10 days – orbiting Sun-like stars of spectral types F, G, and K.

Transit light curve. This diagram illustrates the characteristic dip in brightness caused by a planet in transit across the face of its host star. In the Solar System, we can observe the planets Mercury and Venus transiting across the Sun.

Gravitational microlensing seeks chance line-of-sight relationships between two remote stars located at different distances. Light from the more distant star, known as the source, is bent by the gravitational field of the less distant star, known as the lens. This transient alignment of source and lens – the microlensing event – offers a one-time opportunity for astronomical observations of the lens system at extremely high magnifications (Gaudi 2007). A key advantage of this method is its near-complete randomness, which under the most favorable circumstances enables it to detect planets as small as Earth or as large as Jupiter around any type of star in the galaxy. Key disadvantages are the rarity of microlensing events and the difficulty of following up an initial discovery with more detailed observations. In addition, microlensing observations are typically limited to the host star’s “Einstein radius,” which has inner and outer boundaries of about 1.5 and 4 astronomical units (AU), respectively (Bennett 2008).

So far microlensing has revealed 9 planetary systems, all apparently centered on M or K dwarfs. Appropriately, these two spectral classes account for about 90% of the total stellar population of our galaxy. The detected planets range in mass from Super Earths through gas giants, with proposed semimajor axes between 1 and 7 AU (Bennett 2008, Sumi et al. 2010, EPE). All these systems are quite remote from our own (more than 1000 parsecs/3260 light years), and system parameters typically have large error bars.

Radial velocity or Doppler shift observations capture the elemental spectra of individual stars over time with high precision. Any regular patterns of wobbles that appear in the spectral lines are analyzed as possible gravitational effects of orbiting planets. This method can precisely define an exoplanet’s orbital period, eccentricity, and semimajor axis, but it provides only minimum values for mass. It is best suited to finding Jupiter-sized planets, while remaining insensitive to lightweight rocky planets like Earth, Venus, and Mars. Further, this method is applicable only to main sequence stars less massive than about 1.4 times our Sun, or to suitable subgiant and giant stars. In addition, its effectiveness is typically negated in stars with persistent chromospheric activity and in stars that rotate rapidly (e.g., young stars). Finally, radial velocity observations potentially require a large investment of time and resources, since a candidate planet can be confirmed only after it has been tracked for at least one full orbit. Given these biases and limitations, more than half of all radial velocity detections are planets at least as massive as Uranus traveling on orbits shorter than one Earth year.

Nevertheless, radial velocity observations are the workhorse of exoplanet studies, having characterized more than 400 planets in more than 300 extrasolar systems over the past 15 years. They will continue to be indispensable to planet searches for the foreseeable future. As a result, some acquaintance with the details of this method is necessary to understand the findings of current search programs.

When starlight is analyzed by a spectrograph, it reveals a pattern of vertical lines corresponding to the wavelengths of light that have been absorbed by the target star's constituent molecules. These so-called "absorption lines" in the stellar spectrum can exhibit variations over time, produced by the movement of the star. The term radial velocity signifies the speed at which an object is traveling toward or away from an observer (e.g., one located on or near the Earth) along the line of sight (i.e., radially). Stellar absorption lines shift toward the red end of the spectrum if the radial movement of the star carries it farther away from the Earth, and toward the blue end if the radial movement carries it closer.

A planet accompanying a target star on an approximately edge-on orbit will tug the star away from us when it passes behind the star and toward us when it passes in front. Sufficiently sensitive instruments can measure these minuscule variations, which can then be tracked over time and analyzed for regular patterns that would indicate a planetary orbit. See these examples of radial velocity curves for two different exoplanetary host stars, illustrating the wavelike patterns created by regular variations in each star's motion.

Extrasolar synergies

Apart from the successes or shortcomings of individual detection methods, the combination of various approaches can create a unique synergy that advances our knowledge of exoplanets. The most salient of such potential synergies involves transiting planets. Although in some cases the transit method has obtained detailed measurements of exoplanets that were previously discovered by radial velocity observations, for the past several years dedicated transit searches have first detected a candidate planet and then followed up with spectroscopic analyses. In all cases, a transit detection relies on radial velocity measurements to define key parameters of the target system. This dependency prevents it from operating as a standalone search technique. Nevertheless, combining transit and radial velocity data is the only way to obtain precise values for the mass, radius, and density of an extrasolar planet.

Because the various methods sample distinct regions of planetary systems, they also contribute to a mosaic of large-scale structure. Transit observations reveal the innermost planets in a system, typically within 0.1 AU; radial velocity searches readily detect heavy planets orbiting within a few AU of their stars; microlensing finds planets in the vicinity of the system's ice line, on either side of approximately 2.5 AU; and direct imaging samples the cold outer orbits.

The success of the transit method in particular has inspired a series of robotic space missions dedicated to observing thousands of stars, simultaneously and continuously, for extended periods of time. Early returns from the current Kepler mission, whose field of view includes more than 150,000 stars, suggest that several hundred new transiting systems will be characterized within just a few years (Borucki et al. 2010).

At the same time, the growing numbers of planets observed by radial velocity programs, and to a lesser extent by microlensing consortia, support statistical inferences regarding the frequency and architecture of planetary systems in our region of the Milky Way. One group specializing in high-precision radial velocity searches predicts that at least 30% of nearby Sun-like stars harbor terrestrial or ice giant planets with orbital periods of 50 days or less (Lovis et al. 2009). Taking a larger view, various collaborative groups of microlensing specialists conclude that [1] Super Earth and Neptune-mass planets are more common than gas giants like Jupiter and Saturn in the outer regions of planetary systems (Sumi et al. 2010, Gaudi 2010); [2] giant planets are more common in the region beyond 3 AU (also known as the “snow zone”) than they are at 0.3 AU; [3] our own Solar System is “3 times richer in planets than other stars along the line of sight toward the Galactic bulge;” and [4] only about 15% of planetary systems have architectures like our own, defined as having multiple giant planets in the snow zone (Gould et al. 2010).

Last revised July 2010