How do we know?
Kepler stares at a single spot in the sky, taking many many measurements of the brightnesses of about 100,000 stars all at once. If a planet is circling one of those stars, and its orbit is edge-on to us, then once every orbit the planet passes directly between us and the star. This is like an eclipse, and the light we see from the star drops a little bit. A planet like Jupiter orbiting a star like the Sun will cause a 1% drop in the light we see, because Jupiter has a radius 1/10th of the Sun's, so the surface area of the planet is 1% of that of the star (remember, area = π x the radius squared, so 1/10 x 1/10 = 1/100 or 1%). Therefore the planet blocks 1% of the star's surface, and we see the corresponding drop in starlight.
Earth is smaller than Jupiter, about 1/10th the radius. That means that an Earth orbiting a star blocks 1/10,000th the light of the star, or 0.01%. That's a tiny fraction! From the ground, that's impossible to measure due to fluctuations in Earth's atmosphere changing the amount of light we see from the star. But from space -- hey, that's where Kepler is! -- we can make far more accurate measurements.
And that's what the news is from Kepler. As a test of its abilities, it observed the star known as HAT-P-7, which is known to have a roughly Jupiter-sized world orbiting it every 2.2 days. This planet, called HAT-P-7b, is far too close to the star to be seen directly, but every time it passes in front of the star, the light we see drops. Here's what Kepler saw after observing this system for 10 days:
The top plot shows the data as the planet circles the star. The big dip is due to the planet blocking a fraction of the star's light. The depth of that dip tells us how much of the star was blocked, and therefore the size of the planet. But look along the plot a little bit to the right: see that fainter dip (right under the i in "Magnification")? What's that?
The bottom plot is the same thing but zoomed in to see more detail. That second dip is a lot more obvious. It's not another planet blocking starlight, which is what you might first guess. It's actually the light from the planet being blocked by the star!
The planet is reflecting light from the star, just like the Moon reflects sunlight, allowing us to see it. When the planet passes behind the star, we don't see that light anymore, so the total light from the system drops a wee bit. It's not much, and totally impossible to see from the ground, but Kepler was able to spot it. And that's critical, because it turns out this dip is about the same thing we'd expect to see if a planet the size of the Earth were to pass in front of the star. In other words, the drop in light from a giant planet going behind its star is about the same as we'd expect from a smaller planet passing in front of the star.
The fact that Kepler spied this dip at all means that, if somewhere out there an Earthlike world is orbiting a star, Kepler will be able to detect it!
Another cool thing is in that data too. See how the light slowly rises and drops over time? We're actually detecting the phases of the planet as it orbits the star! As the Moon orbits the Earth, we see it going from mostly dark (new Moon) to half lit to full, getting brighter over two weeks. Then once it's past full we see more and more unlit surface, so it appears to dim over time. The same thing is happening to the planet HAT-P-7b as it orbits the star. Right after the eclipse event we are seeing it as "new", with the dark side facing us. As it orbits, we see more and more lit up, until it passes behind the star. After that secondary eclipse, we see the light from the planet get dimmer. The Kepler website has a great animation showing this, which I also uploaded to YouTube:
What this means is that the Kepler data are showing us the phases of an unseen planet orbiting a star more than a thousand light years away!
There is some bad news, sadly: if we want to find an actual Earthlike planet we have to be patient. It takes the Earth a year to orbit the Sun, right? So suppose we see a small dip in the light from some star that indicates a planet the size of the Earth is orbiting it. To confirm that, we'd have to wait a year to see that dip again! And even that's not enough, since we don't know beforehand how long the orbital period of the planet is. That second dip might be from sunspots, or another planet, or something else entirely. So we have to wait again, and spot the dip a third time. If the time interval between the two dips is the same, then we can be pretty sure we're seeing a small planet eclipsing (actually, the correct word is transiting) its star.
In general the orbit could be days long, as it is for HAT-P-7b, or it might take months or even a year. And remember the real prize here is to find a planet like Earth, which means an orbit that takes months or more. So we really won't have those kinds of results from Kepler for a while yet. It's only been up and observing for a few weeks.
But come 2011 or 2012, and we may have our answer. Imagine! In just two years we may know if other Earthlike planets are orbiting stars in our galaxy!
Most astronomers, including me, assume that these planets exist, but it'll be incredible to have the actual data. And better, Kepler is looking at so many stars that we'll get actual statistics for these planets. We'll be able to guess just how many such planets exist! Are we a rare specimen, or are there millions of Earths out there in the galaxy?
At this moment, this exact moment in history, the Earth is a lone habitable rock orbiting one star in the depths of space. But in just two more of our own orbits, we may suddenly find ourselves located in a planetary metropolis studded with vast numbers of worlds, all of which were just waiting to be found.