So astronomers do what they can to understand this menace from space. We look for rocks on orbits that intersect ours, we think about ways of moving them out of the way should we find one, and we also think about the record we do have of past impacts to see what we can learn from them. There are about 180 impact craters known on our planet, ranging from tens of millennia in age to billions of years. They also vary in size from a few kilometers across to monsters so big they can only be detected from space. Sometimes it's hard to measure their size (they can have multiple concentric rings, or be underground -- covered up due to extreme age -- making definite sizes hard to figure out) or hard to get their age. But we do have some statistics on them, and there have been many studies about them. A big question is: are impacts periodic? That is, do they happen with some repeating period? If so, then there must be some astrophysical cause: a giant planet in the outer solar system, for example, that shakes loose comets every 50 million years, or the Sun passing near another star. This has been studied, and all kinds of periods have been found in the data. I've always been a little skeptical of them, since the data are sparse. And now it looks like my thoughts are being supported: a new study finds no such pattern in the ages of craters, and concludes all the periods found previously are probably due to errors in the analyses. The difference is that the author, Coryn Bailer-Jones, used Bayesian statistical methods. This is different than standard statistics, and is less prone to bias due to uncertainties in age and size of craters. In using standard statistics, clusters in crater ages can always be found, but it's hard to know if that's just a random clump or has an actual physical cause -- like flipping a coin 10 times and having it come up heads 5 times in a row. It's unlikely, but how do you know if it's coincidence or not? Bayesian methods circumvent that issue. Bailer-Jones found that when the uncertainties are treated properly, the periodicity of impact events seen in previous studies vanishes. We're not bombarded every 60 million years (or whatever); it was simply an artifact of the way the math was being done. And while I won't go so far to say that this closes the book on periodic impacts, it makes it look a lot less likely*.
There were two other results found that I think are particularly interesting, too. One is that with huge craters, ones bigger than 35 km (21 miles) across -- in other words, more likely to cause extinction-level events -- there is no trend in the rate of impacts over the past 400 million years. That is, we're getting hit just as often now as we were back then, which is to say, not very often. The second is even more interesting: for craters of all sizes (not just big ones), there is a trend of increasing rate over the past 250 million years. Are we getting hit more by smaller objects now than we were in the past? Not necessarily! That's one explanation, of course, but another is that smaller craters are more easily eroded over time. We don't see a lot of old ones because they fade away due to wind, water, earthquakes, and so on. That makes it look like we're seeing more now, but that may be an illusion; once erosion is taken into account the rate of impacts looks like it's steady once again. However, I'll note that when you look at the Moon there's evidence that impacts have increased over time, so these conclusions are by no means firm. After all, we're basing them on a relatively small number of craters, but it's what we're stuck with. So it looks like impact rates from space are not periodic, at least not over the past 400 million years or so. We're not "due" for one any time soon. And perhaps more importantly, given how some media tend to report these things (cough cough), there's no solid evidence that we're more at risk now than sometime in the past. I know a lot of people worry about these things, so I want to be clear on that. And the good news is, as I pointed out recently, when their number came up dinosaurs didn't have a choice to live or die. We have a space program, so the choice is ours.
^* Note that this study was only using impact from space. There are studies that do show periodic rates of mass extinction on Earth. If true, these are due to other things, and not impacts. One interesting event may be due to the Sun's orbit around the Milky Way. I find this idea compelling, though the evidence is still too sparse to be reliable.
Image credit: cartoon:Blastr; Barringer crater in Arizona: National Map Seamless Viewer/US Geological Service
Related posts: - New meteor shower points to a future close encounter - No, 2005 YU55 won't destroy the Earth - I think the Moon watched Bad Universe - Repeat after me: Apophis is not a danger (and note this followup, too)