Six ways to find another Earth


The 2004 transit of Venus across the Sun, as viewed by NASA's Transition Region and Coronal Explorer (TRACE) satellite. The faint red ring around Venus is a consequence of sunlight scattering off its atmosphere.

When you picture an astronomer looking for planets around other stars, what do you see?

I'm guessing many of you are summoning images of Galileo Galilei sketching on parchment and gazing upward through a hand-held spyglass, or Edwin Hubble sitting all night in a cavernous, cold observatory dome, pipe clenched in teeth, peering through the eyepiece of a gargantuan telescope and exposing photographic plates. But that's rather like imagining a dentist extracting a tooth with string and a strongly slammed door, or a mechanical engineer performing calculations with an abacus. These days, rare is the professional astronomer who actually looks through a telescope at all, and rarer still those who make profound discoveries doing so. The human eye simply isn't up to the task of discerning subtle details—and the human hippocampus is inadequate for recording them.

This is part of the reason why most exoplanets don't swim fully-formed into view through a telescope's lens so much as they gradually take shape from a mist of statistics and inference. But an even bigger contributor to this observational dislocation is that, in comparison to stars and the vast distances between them, planets are so faint and small that actually seeing them at all is incredibly challenging. Of the six main methods for detecting exoplanets, only one, direct imaging, hews to the antiquated notion of observing and studying an object by taking its picture. The other five—astrometry, radial-velocity spectroscopy, microlensing, transits, and timing—discover planets by observing stars and searching for the subtle effects induced by any accompanying worlds.

You'd think that direct imaging would be the easiest to understand and explain, but it actually requires so much complex, cutting-edge, and just plain cool technology that I'm saving it for another post. Direct imaging using specialized space telescopes also deserves separate discussion because it may be the only way to deliver the results everyone wants: compelling evidence of a complex biosphere on an Earth-sized planet orbiting another Sun-like star. For now, suffice to say that both ground- and space-based direct-imaging techniques can only directly image very large, hot exoplanets orbiting far from their stars.

The remainder of this post will discuss the other five techniques, all of which hold promise for discovering potentially habitable worlds, but have somewhat less to offer in terms of deeply investigating them. They are perhaps less glamorous, but they form the foundation of planet hunting, and are crucial for understanding all the exciting discoveries that will be announced in coming years, months, and days. Let's get started, there's a lot of ground to cover.

Wibbles and Wobbles

Two of the methods, astrometry and radial-velocity spectroscopy, look for "wobbles" in a star's motion caused by the gravitational to-and-fro tug of orbiting planets. A wobble's periodicity—how often it repeats—indicates a planet's orbital period. And a wobble's strength—whether it corresponds to a kilometer or a centimeter of motion, for instance—indicates a planet's mass.

Astrometry looks for wobbles by measuring a star's motion within the two-dimensional plane of the sky, as judged against the positions of relatively static, more distant stars. So if, from our perspective on Earth, a star is orbited by a planet that tugs it to the right and left, or up and down, the astrometric signal is detectable.

Astrometry is actually the oldest planet-hunting technique—it dates back to the mid-19th century. But it's also the most underutilized, perhaps because all of the planets it has discovered have later proved illusory. See, for instance, the story of Peter van de Kamp, or the recent retracted astrometric detection of the exoplanet VB 10 b. To detect the astrometric signal of an Earth twin around a nearby Sun-like star, you'd need a measurement precision of something like one-millionth of an arcsecond, which, if my math is right, is akin to discerning the thickness of a dime from more than half a million kilometers away. Astronomers have plans for spacecraft that could perform such precise measurements, but currently lack the funding to build them.

If astrometry is the overlooked and difficult wallflower of planet hunting, then its sibling, radial-velocity spectroscopy (RV), is the all-star varsity quarterback who makes straight A's and dates the prom queen. As of this fine February 2nd morning, RV has detected more exoplanets than any other technique, but that status may change later today. Unlike astrometry, RV detects wobbles perpendicular to the plane of the sky—in other words, it detects motion toward and away from us here on Earth, not back and forth in the sky.

It does this spectroscopically, by separating a star's light into its constituent colors, just like how sunlight becomes a rainbow when fed through a prism. When a planet tugs its star toward us, the starlight shifts toward the blue end of the spectrum; when a planet tugs its star away, the starlight shifts toward the red. The effect is exactly the same as when the sound of an ambulance siren increases in pitch as it approaches, then decreases in pitch as it whizzes away. Interestingly, this means that the ideal exoplanetary system for RV detection, in which the planet's orbits are inclined edge-on to Earth, would be invisible to astrometry. And vice-versa: An ideal astrometric planetary system, with its orbits aligned precisely within the plane of the sky, would be invisible to RV.

RV most easily detects massive planets in scorchingly close orbits of their stars, which is why most of the exoplanets found by early RV searches were so-called "hot Jupiters." Our Earth's motion around the Sun causes our star to shift its position by 1 centimeter per second over the course of a year. That wee signal is what RV searches for alien Earths have to detect across the light-years. Ground-based RV surveys aren't quite reaching that degree of precision yet, but they're getting very close, primarily by combining years of observations to amplify faint signals.

The biggest downside to RV as a technique is probably that, because it can't reliably measure the inclination of an exoplanet's orbit, it can generally only provide an estimate of a planet's mass with a fudge factor of around 15 percent. That may sound small, but it could be the difference between a planet being a comfortable place to live or a frozen, desiccated chunk of lifeless rock. Astrometry provides a true measurement of mass, but is rather more difficult to perform.

Quantity and Quality

Beyond the "wobble" methods, microlensing and transits use photometry—measuring the quantity of starlight—to find planets.

Microlensing is unabashedly weird, and rather counter-intuitive, a product of living in a universe that operates using the relativistic rules of Einstein. General relativity dictates that the more massive an object is, the more spacetime curves around it. Very massive objects like galaxies can curve spacetime so much that they act as gravitational lenses, amplifying light from background objects. Astronomers reserve the term "microlensing" for stars, which create correspondingly smaller gravitational lenses. If a foreground star with accompanying planets passes in front of a far more distant, precisely aligned background star, the gravitational fields of the planets can further enhance the lensing star's magnification effect. Astronomers detect this as a transient brightening in the lensing event over a period of hours or days, and can typically obtain a lensing planet's mass and orbital separation.

Microlensing alignments are rare, so astronomers must monitor very large numbers of stars (for instance, those in the galactic bulge) to have a reasonable chance of detecting events. But the technique is very powerful: Space-based microlensing observations can easily reveal the presence of Earth-mass planets in the habitable zones of stars.

Unfortunately, each lensing event occurs only once, and typically reveals planets many thousands of light-years away from our solar system. This means that any tantalizing planets discovered via microlensing will almost certainly be beyond the reach of meaningful follow-up observations. Consequently the technique is most useful for studying the architecture of planetary systems and the statistical distribution of planetary masses and orbital separations. For anyone hoping to find evidence of life on other planets, microlensing-based detections are a little unsatisfying.

Transits, on the other hand, are the jackpots of planet-hunting. A transit occurs when a planet crosses the face of its host star as seen from our solar system, fractionally diminishing the star's light as it passes. Transits are, by their nature, rare occurrences, since for any particular line-of-sight the chance of a precise planet-star alignment is small, dictated by the ratio of the diameter of the host star to the diameter of the planet's orbit. Thus, larger planets on short-period orbits of small stars are more likely to transit, and large numbers of stars must be surveyed for any transits to be found.

Since they are silhouettes, transits yield a planet's diameter, and reveal a planet's orbital period by their recurrence. However, follow-up measurements using the RV technique can usually give an estimate of a transiting planet's mass, provided the planet is within several hundred light-years of Earth. Pairing the planet's mass and the diameter yields its density, which helps astronomers pin down whether a planet is, for instance, made mostly of gas or of rock.

Sometimes, astronomers can even study a transiting planet's atmosphere as starlight filters through or reflects off it at key points in the planet's orbit. In this way, planet-hunters can obtain information about a transiting planet's atmospheric composition, its temperature, and even what sort of weather it has. Such measurements have already been performed for very large transiting planets around nearby stars. NASA's James Webb Space Telescope, launching no earlier than 2015, might be able to perform similar studies for a handful of smaller, more Earth-like planets that transit small, dim, red stars called M-dwarfs.

Seen from interstellar distances, the transiting Earth would dim the Sun's light by one part in 12,000; that's less than a current-generation iPad dims when it shorts a single pixel that's like detecting a clump of 65 dead pixels out of the nearly 800,000 on a current-generation iPad display. Yet astronomers have already built space telescopes capable of detecting such minute changes in starlight: NASA's Kepler mission uses transits in its 3-year quest to pin down the frequency of Earth-size planets in our galaxy. The 3-year running time is key, as this allows a transit of an Earth-size planet in a yearly orbit of a star to recur 2 or 3 times, ensuring that the diminutive dip in starlight is actually caused by a planet. We'll probably be talking a lot more about Kepler throughout the remainder of my guestblogging tenure at BoingBoing, because the mission is slated to release its next treasure trove of discoveries later today.

The final technique, timing, is really a grab-bag of methods that all use gravity-induced variations in the timing of various astrophysical phenomena to indirectly detect planets. The first exoplanets to be discovered were found by timing the minuscule offset their masses caused on the clockwork regularity of the rotations of neutron stars. Others have been found by the timing variations they induce in the pulsations of the outer layers of stars.

But timing methods really come into their own when they are applied to a transiting planet: Subtle variations in the periodicity or duration of a planet's transit can reveal the gravitational influence of otherwise-unseen companions, potentially even the presence of large accompanying moons. And given that the Kepler mission has the sensitivity and precision to detect such variations in its probable yield of many hundreds of transiting exoplanets, transit-timing appears poised to provide several notable discoveries in the near future.

If you made it this far, congratulations, you're now pretty well-prepared to grok whatever exoplanet-related finds and claims you'll encounter over the next few years—and the next few days, for that matter. The Kepler data release is almost upon us, and I guarantee it will be filled with wonderful surprises.

Want more? In addition to what I've written here, I'd direct you to the Wikipedia page on exoplanet detection, which is extremely comprehensive, as well as a 'Cribsheet' I produced for Seed magazine that visualizes the basics of three techniques. You can also see the tally of planets that have accrued to each technique in the chart I posted Monday.


  1. Great article Lee. Love a good hugeness of space analogy. Look out for an interview with Sara Seager of MIT later today @theurbantimes.

  2. “Of the six main methods for detecting exoplanets, only one, direct imaging, hews to the antiquated notion of observing and studying an object by taking its picture.”

    “Antiquated”? Doesn’t NASA spend billions on projects that directly image space? Isn’t this how objects (comets, asteroids, galaxies etc) continue to be found today?

    1. I presume that the terms “retro” and “antiquated” were used in their most accurate sense: visual identification is the very oldest form of astronomical observation. I grant that words with less negative connotations might have been used, but I thought it was a brilliant and thoughtful post.

      I am looking forward to the Kepler data release!

    2. Finding comets, asteroids and galaxies is very different from finding exoplanets. Looking for an exoplanet next to its home star is like looking for a firefly against a klieg light. The star’s light entirely dominates.

      One clever way to work around this is to create an artificial eclipse of the star. Take the light from the star. Feed it through some optics so that the light wave is flipped upside down–what was once a crest is a trough, and vice versa. Now combine the upside-down light wave with the right-side-up wave, and voila, they cancel each other out. All that’s hopefully left is the firefly-level light from any planets. That’s a really simplified version of events, and actually making the thing work is a lot more time and money. Yet not impossible!

      1. I didn’t say that they were the same. In fact I never mentioned exoplanets in my post. I merely took exception to describing the technique as “antiquated” when it is incredibly successful in other areas. It’s a minor point and I don’t think the author intended it that way but it implies obsolescence.

    3. A curious thing, though, is that direct imaging doesn’t seem to tell you very much. At least for the first planet found in visible light, Fomalhaut b, most about what we know about it comes from properties of the surrounding disk. Whereas other techniques can at least give you a good idea of the planet’s orbit and mass.

  3. The statement:

    “…Earth would dim the Sun’s light by one part in 12,000; that’s less than a current-generation iPad dims when it shorts a single pixel.”

    This seems incorrect to me. The iPad’s screen is 1024 x 768 = 786,432 pixels. If we assume that we start with all pixels lit, turning one off would dim the screen by a factor of 1 part in 786,432. In other words, I can turn one iPad pixel off every second and it would take a bit over 9 days to go completely dark, whereas if I blocked 1 (of 12,000) of the Sun’s “pixels” every second, it would appear to go dark in 3 hours and 30 minutes or so.

    1. Egads, Craig, you’re totally right. This is why I prefer writing over calculating. I take pains to curb my innumeracy but this is a flagrant, embarrassing error. I backtracked and found what I did wrong in my calculations… Several things, actually. Let me see if I can get this post corrected.

    2. Okay, so it looks like the 1/12000th dimming would actually correspond to about 65 pixels droppings out from the iPad display… Right?

  4. 65 works for me. And 65 dead pixels would be really EASY to spot for the human eye, so it sounds unimpressive. But hold it far enough that it’s just a point of light, and have it display an all white web page, apart from the word “ipad” in about 12 point text, in a blink tag. I doubt the naked eye could spot the change in brightness with the blinking (and not just because no self-respecting browser respects that tag).

    If you don’t want to be parted from your ipad from such a distance, shine the light from the ipad screen on the wall, and try to tell the difference in illumination – similar concept.

Comments are closed.