Was the fourth time the charm for Kepler?
No—NASA rejected us again. This time they said we needed to show that even in orbit, where it is dealing with cosmic rays and other noise, the telescope would be sensitive enough to detect Earth-size planets. At least they gave us $500,000 to demonstrate this in the lab. Ames came in and lent us another $500,000. NASA headquarters thought it would take us two or three years to design and build this thing. Then we’d miss the next chance for a proposal and go away and get out of their hair.
This time you had to build a prototype yourself, with no “procurer” to help. How did you do it?
We built a closet-size device, about 4 feet by 4 feet and 10 feet tall. At the bottom we placed a globe and shined a light in it. The light came out of the globe and through a metallic disk with laser-drilled holes. Each hole represented a star. At the top we put a little telescope and a ccd to detect the brightness of all the stars. Then we jiggled the telescope just like what would happen in space. The hardest part was simulating dips in brightness of 84 parts per million, the amount of dimming caused by an Earth-size planet around a sunlike star. If you just stick a piece of glass in front of one of the holes, it reduces the brightness by 40,000 parts per million. So we had to invent something new. We put very fine wires across some of the holes, and then ran current through the wires. When the wires got hot, they expanded by the diameter of a few atoms and blocked just the right amount of light. With all that in place, we showed that even with vibrations in space, the detectors could make these observations.
NASA figured this would take you a few years. How long did it take for real?
We got it to work in six months, sent a technical report to headquarters, and submitted our proposal on time. In January 2000 NASA finally said, “We give up. Here’s your money.” They gave us some funding but told us we were competing against two other missions. Then in 2001 they officially approved us and set a launch date of 2006.
But then you ran into money problems.
Right after we got funded, NASA told us we weren’t going to get any money the first year. I was outraged. We had critical people who would leave. Fortunately, Anne Kinney, who was a director of science missions at NASA headquarters, acquired money—from various places we probably don’t want to go into—and gave us $1 million to get started. We ran into another problem in 2005, while we were building things rapidly. NASA informed us six months into the fiscal year that they had cut our budget in half, which means we’d already spent everything we were going to get for the year. Our only option was to fire everybody. Then the next year we had to hire new people and retrain them. That increased the total cost of the mission by many millions of dollars and caused a year of delays.
Despite all that, you had Kepler ready to go in 2009. How was the lead-up to the launch?
Kepler was set to launch on a Delta II rocket, and those rockets were having trouble at the third stage. The fuse didn’t work well. Sometimes it would light at the wrong time or not at all. Right before our launch, there were two new Deltas sitting on the pads at Cape Canaveral, one for a military gps satellite system and one for Kepler. The military told us, “You get to launch first. We want to see if the fuse works.”
That must have made you extra nervous.
There was some anxiety. But after that many years, you’re pretty used to anxiety.
What was it like watching Kepler finally make it into space after all those years of planning?
It was wonderful just to see Kepler go into orbit. But one of the biggest excitements came three or four weeks later, when we saw the full-frame image come in. All the stars were there. The spacecraft was not jiggling. It was just a marvelous image.
OK, time for the nuts and bolts. How does Kepler actually work, hunting for planets?
Kepler is like a giant camcorder constantly taking images of a group of about 170,000 stars in one patch of sky. The system takes an image every six seconds and then records the 30-minute average brightness for each star. We store all that data on the spacecraft for a month. Then the spacecraft rotates toward Earth and transmits the data to us.
That sounds like a bewildering amount of information. How do you make sense of it?
We analyze the data once every three months. There’s a lot we have to account for: Star brightness readings are affected by temperature changes on the spacecraft and the electronics, and some stars change brightness naturally. So we have to measure all those different processes and cancel them out. Only then can we look for transits, those little brightness dips that indicate planets. We have a computer program search through all the data looking for little repeating dips; repeating means that you are seeing the same planet going around and around. If the program finds at least three potential transits around a particular star at regular intervals, then it flags that star for follow-up.
How many new planets has Kepler discovered around the 170,000 stars you’re watching?
The computer has flagged about 15,000 possible planets, but a lot of those are false positives. For example, there are lots of stars that cross in front of each other, which causes a similar dip in brightness. So we get a group of scientists together at Ames to do what we call triage. They glance through all the crossing events and identify the ones that look promising. These are known as “objects of interest,” and they number about 4,000. Then we have another team that looks through this data in a huge amount of detail. If the object passes that test, then we upgrade it to “planetary candidate.” We have identified more than 3,000 candidates, and we believe that most of them really are planets. But we don’t call any of them planets until observers can confirm them.
How do astronomers confirm a planet around another star trillions of miles away?
They use ground telescopes to look for the gravitational tug the planet would make on its star. They use space observatories like the infrared Spitzer telescope to confirm the transits. So we execute a whole stack of ground- and space-based measurements, along with some very clever analysis of the data. Only if a candidate passes all those tests do we say it’s a confirmed planet. Once we say we’ve discovered a planet, we will bet our careers that it is a planet. It’s not a false positive. If we say it’s an Earth-size planet in the habitable zone, there is no way mankind knows of anything more we can do to prove it really is a planet.
What does Kepler tell us about the overall number and variety of planets out there?
Based on Kepler’s growing planetary candidate list, it is clear that our galaxy contains at least 150 billion planets, and that at least half of its stars have planets. The 3,000 candidates range from well below the size of Mars to nearly double the size of Jupiter. About 1,500 are twice the size of Earth or smaller, meaning they might very well be rocky planets. Most of them are like Neptune: ice giants with lots of rock and ice but also hydrogen and helium atmospheres inhospitable to life. However, we expect that many of these Neptunes have moons. There are several groups looking for these moons, since some could be Earth-size and temperate enough for life.
What are your favorite days working on Kepler?
The highlights are the days when I see the rough drafts of papers with the proof—not the speculation, not the hope—that these planets exist. I’ve read things like “We found a planet orbiting a binary star.” For a long time people didn’t believe there could be planets orbiting a pair of stars. We proved they do. I’ve read papers that found small planets with a high density. For sure those are rocky planets, like Earth or Venus. Those papers get me really excited.