Kamuela, Hawaii - It doesn't seem like a place where planets are discovered, only steps away from a roaring four-lane highway, a congested shopping center and a McDonald's.
But this is the planet-hunting base for astronomers Geoff Marcy of the University of California at Berkeley and Paul Butler of the Carnegie Institution of Washington, D.C. They are the world's most prolific planet finders and do much of their work at the W.M. Keck Observatory headquarters, here in the shadow of the mountain where its twin 10-meter telescopes, the world's largest, reign supreme.
This month, they'll begin a new survey of 400 stars from the Magellan telescope in Chile, searching for planets that can only be seen from the southern hemisphere, Marcy said.
It will be part of their continuing their survey of the nearest 2,000 stars, which will help NASA choose suitable targets for space-based telescopes in its long term quest to find an Earth-like planet. Their search, which began in 1986, will occupy their time for the next 10 years.
Marcy, Butler and their team, the California and Carnegie Planet Search, have found about two-thirds of the 101 planets known to exist outside Earth's solar system. Their work is funded by the National Science Foundation, NASA and Sun Microsystems.
"What they have done is extremely profound, but not surprising," said University of Wisconsin-Madison astronomy professor Robert Mathieu.
He said the surprise came in the nature of the planets they found, such as "hot Jupiters" orbiting close to their stars. All planets detected so far are gas giants, many of them much larger than Jupiter and all much larger than Earth.
"Their experimental method was superb," said Mathieu, who is not involved in their planet search. He said when they announced their findings, "There was not much argument. This does not happen often in science."
Marcy and Butler don't actually see the planets. Instead, they watch for evidence of the subtle tug a large planet makes on its star. They do this by examining the stellar spectrum, night after night, to watch for minute changes in the star's color.
The color change reveals stellar motion, called radial velocity. The technique is so refined that they can observe a star's velocity vary by only a few meters per second, the speed of a slow jogger.
Many variables must be factored before any conclusions are reached. Butler says it took him eight years to perfect the software that analyzes the enormously complex data they gather on each star.
In the late 1980s, when their planet search began, "it took approximately six hours of computer time to reduce a single observation," Butler said.
It now takes about 10 minutes. Back then, he'd run his programs in the backgroundof his colleagues' computers. It got so bad that one institution threatened to take away his computer privileges.
To detect this tiny change in a star's movement, Marcy and Butler exploit the Doppler shift. If a star is moving toward the Earth, its light turns slightly blue. If it's traveling away, its light shifts to red.
The trick is to have a spectrometer accurate enough to detect a shift as small as one part in 100 million. This was not possible with equipment available in the 1980s, so the two developed a better spectrometer.
They collaborated with astronomer Steven Vogt of the University of California at Santa Cruz. "He built the greatest spectrometer in history," Butler said.
A spectrometer is like a refined prism, spreading starlight into a rainbow. Looking closely at this rainbow, an observer sees dark lines, a result of elements in a star's atmosphere absorbing specific wavelengths of light.
Each element leaves a unique dark-line signature. When the star is receding, the lines shift toward the red. When coming toward Earth, the shift is blue. How much shift depends on how fast the star is moving relative to the observer. Marcy, Butler and Vogt learned to measure this shift hundreds of times better than anyone before them.
That precision is due in part to Butler's development of a glass flask that holds iodine gas. The starlight passes through this flask, which is placed inside the spectrometer, and allows the team to compare the starlight spectrum to the iodine spectrum. The flask is crucial to the accuracy of their measurements.
Just before Halloween, Marcy, Butler and graduate student John Johnson have the Keck I remote observing room to themselves.
It's a comfortable office lined with computers, and they're in comfortable T-shirts and shorts.
They communicate with telescope operator Julie Rivera by a televised Internet hookup. The telescopes are 20 miles away. Rivera is at the top of the mountain, in the thin air of Mauna Kea. She has their list of targets, and they start to work immediately, taking spectroscopic pictures of 103 stars. It will take all night.
During a break, Marcy describes one star they are observing, called Gliese 876, in the constellation Aquarius about 15 light years away. It can't be seen with the naked eye. This star has two planets in orbits of about 30 and 61 days.
"They are in resonance," he says. "We are now seeing changes occurring in the planet's orbits. We can detect the perturbations they have on each other," he says, which gives them more clues to a planet's mass.
This is important to planet hunters, because the Doppler detection method can only tell them a planet's lowest possible mass.
But there are other ways to detect planets.
A planet can pass in front of a star, slightly dimming it for a brief period. This is called transit photometry.
Greg Henry of the University of Tennessee confirmed the existence of a planet orbiting the star HD209458 in the constellation Pegasus, using this technique. He saw a slight dip in brightness when the planet passed in front of the star. He used a robotic telescope in the Arizona desert, which tirelessly scans the heavens, recording and analyzing data with no human attendant. The finding was significant because it independently verified the technique used by Marcy and Butler.
A planet also can act as a gravational lens, making star light brighter for a brief period. But the chances of a planet passing between its star and our line of sight are quite low. "One in a thousand," Marcy said.
And so far, with more than 1,300 stars surveyed, this remains the only confirmed transit. But the transit method shows promise, and many other teams are using automated telescopes and powerful computers to look for evidence of transiting planets.
David Charbonneau of the California Institute of Technology is using off-the-shelf telephoto lenses, custom digital cameras and complex software to search for planetary transits from the Palomar Observatory near San Diego.
His cameras photograph thousands of stars at a time and computers compare images taken at different times to see if any of the stars show signs of a transiting planet. This system uses numbers to work in his favor. It can take 300 pictures each night, with an average of 20 cloud free nights per month. That makes 6,000 images, each containing thousands of stars, for the computers to analyze each month.
While many planets have been detected, none has been seen yet. This will change, said Fred Chaffee, director of Keck Observatory.
"It's going to happen in the next two years. We have the technology," he said.
Getting an image of a planet, Chaffee said, "is the current low-hanging fruit on the planet-finding tree."
A team headed by Ben Zuckerman at the University of California-Los Angeles is looking closely at nearby young stars in hopes of visually detecting a planet.
Denise Kaisler, a PhD candidate and a member of the team, described her team's search.
"We've observed about 100 stars at 50 parsecs and closer since February 2000. All have been at the Keck, using adaptive optics." A parsec is about 3.26 light years.
"No one, to our knowledge, has gotten an image of a planet near a star. What we are doing is totally new."
But the stars are more than a billion times brighter, she says, in visible light. To solve this problem, Kaisler's team observes a select group of young stars in infrared light. She explained that planets in young solar systems are relatively brighter at this wavelength. The team also uses adaptive optics, and the combination of adaptive optics and observing in infrared can concentrate the light of such a planet until it is only a million times fainter than its parent star, a thousandfold improvement.
Astronomers have tried to see the wobble a planet makes on star's position by tracking its position relative to distant background stars, but with no luck to date. In 1963, Dutch-American astronomer Peter van de Kamp said a Jupiter-sized planet orbited Barnard's star, 5.9 light years away. He based his claims on astrometric detection, but his data was disputed.
Astronomer Jennifer Bartlett of the University of Virginia analyzed his data, made more observations and reported in June 2001 that Van de Kamp's claim could not be supported. Some speculate that flaws in his telescope were not taken into account.
At this point, finding a planet based on astrometric detection shows promise with a NASA space-based telescope called the Space Interferometry Mission. It is being designed to determine stellar positions to one-thousandth of an arc second, or about the length of a yardstick on the moon, as seen from the earth. Launch is planned for 2009.
For now, the detection of an Earth-sized planet is not possible. Even the Webb Space Telescope, which will replace Hubble, will not be able to do it.
But Marcy points out that they can find a solar system with a "Jupiter-analog," meaning a solar system with a Jupiter-sized planet at a Jupiter-like distance, in an orbit that is roughly circular.
"That will be a signpost," he says, for what is the holy grail in the planet-finding community - "a rocky planet with liquid water."