How does carbon dating work?

We know the number of particles that are present when a radioactive element is complete. It decays at a constant rate – that is, it loses half of it energy (mass) over a fixed period of time. We call this the “half-life” of the element. If it has a half-life of 10,000 years then after 5000 years it has half the number of particles left. Then after 2500 years it would have ¼ of the original left and so on. The environment – temperature, soil erosion or any other factor has no effect on the rate of decay. Some elements have very long half-lives and others very short ones. However they can all be seen as types of “atomic clocks” as their decay (loss of energy or mass) is constant and we can work back to see how old something is. Carbon dating is just one form of radioactive decay and is used to date organic material. Other elements date non organic material. Sometimes called “clocks in the rocks.” As mentioned above the Dawkins books has a very good chapter on it.

Good job, Nelson!

Sophie,

One thing to be aware of is that many people use "carbon dating" to mean any sort of radiometric dating, not just the dating of recent organic samples.  (I even heard them make that goof on "Universe" early in the first season.)  This is unfortunately because carbon dating has a lot of very important differences from other variants, not just in terms of how far back one can date, or what it is being dated (plant or animal matter vs. rock), but because carbon dating is the only method that has to be calibrated.  Since the carbon-14 decays to nitrogen-14 which is already butt-common in organic matter, we have to measure the ratio of c-14 to c-12, but just doing that can be misleading without calibration because the starting ratio varies over time.  Calibration is just the process of accounting for that difference. 

That's an added extra complication that creationists will try to exploit to make the whole thing sound very iffy.  Which is ironic because most of the creationist disputes are with ages we got with other methods that don't have this issue at all!  Carbon dating of course will never be used to give evidence that the earth is 4.5 billion years old, but other methods are more useful and far simpler to deal with.

Nelson mentioned potassium-argon dating but there are many other possible methods and one commonly mentioned is uranium-lead, and I am a bit more familiar with it than potassium argon.  Uranium has a 4.5 billion year half life, so it's eminently usable for any rock we will find on earth.  Take a big hunk of granite.  Find a zircon crystal somewhere in it, there probably will be some, very tiny ones.  When zircon crystallizes, it can pick up uranium impurities but not lead impurities.  Thus the crystal starts out with no lead whatsoever.  [Creationists who are smart enough to know you don't use carbon dating to date trilobites, but are still stupid ignorant enough to be creationists, will try to claim that we have no way of knowing whether or not there was any lead to start with in the rock.  But as you can see this is simply bullshit, and the ones who work at the discovery institute or the creation science institute and write all the propaganda the less educated latch onto, goddamn fucking well know enough to know this, the lying sacks of shit.  Not that I am bitter.]  So any lead found in the zircon today was formed from uranium decay, decay of uranium that got locked into the crystal when it solidified.  To avoid any possible chance of contamination, though, they remove the outer layers of the zircon crystal before they start.  The beauty of the uranium-lead test though is that it is really TWO tests; there are two different isotopes of Uranium, 238 (with a 4.5 billion year half life) which eventually decays to lead 206, and 235 (with a 700 million year half life) which eventually decays to lead 207.  You can run both tests, and see if they each give the same answer for the age of the zircon crystal.

As for precision, nowadays they don't weigh the samples, they run them through a mass spectrometer (a lot of old cyclotrons that aren't powerful enough to do any new work in nuclear or particle physics are doing this work in retirement) and can count individual atoms.  When I was a kid, the Cambrian was thought to have begun 560 million years ago.  Now with more accurate measurement, they say it was 542 million years ago; imagine having it down to two tenths of a percent uncertainty!  [Of course, what this really means is the evil-lutionary scientists can't make up their minds what lie they want to tell!  NOT!]