temperature reconstruction in ice cores

(picture from Danish Center for Ice and Climate)

So, how do scientists figure out the temperature of the past by looking at those ice cores? The not so simple (but really interesting!) answer:

the oxygen (O) in these cores has both 18O and 16O isotopes present. 18O is heavier than 16O (because it has two more neutrons, but all other properties are the same). By looking at the amount of each of these isotopes two scientists can determine temperature. Graphs usually will show δ18O, which means the amount of the heavier of the two isotopes. Low δ18O is associated with low temperatures, because...
So, this is generally what happens during an ice age. There are both 18O and 16O isotopes in the ocean. More 16O evaporates out, since it is lighter in weight. So more 16O ends up in the atmosphere, and more 16O, less 18O precipitates down as snow onto the ice sheet. That means that the ice ends up containing more 16O than 18O total since there is no melt off from the ice to the ocean during an ice age, the isotopes are locked into the ice. During a warm period, there is melt off to the ocean, so the ratio is more balanced. So, when we now look at ice cores, if there is relatively not much 18O locked in the ice, we know that the earth was at a colder temperature. Yay science!