One of the best ways to measure farther back in time than tree rings is by using the seasonal variations in polar ice from Greenland and Antarctica.
Ice cores are obtained by drilling very deep holes in the ice caps on Greenland and Antarctica with specialized drilling rigs. As the rigs drill down, the drill bits cut around a portion of the ice, capturing a long undisturbed core in the process. These cores are carefully brought back to the surface in sections, where they are catalogued, and taken to research laboratories under refrigeration. A very large amount of work has been done on several deep ice cores up to 9,000 feet in depth. Several hundred thousand measurements are sometimes made for a single technique on a single ice core.
There are a number of differences between snow layers made in winter and those made in spring, summer, and fall. These seasonal layers can be counted just like tree rings. The seasonal differences consist of
a) visual differences caused by increased bubbles and larger crystal size from summer ice compared to winter ice,
b) dust layers deposited each summer,
c) nitric acid concentrations, measured by electrical conductivity of the ice,
d) chemistry of contaminants in the ice, and
e) seasonal variations in the relative amounts of heavy hydrogen (deuterium) and heavy oxygen (oxygen-18) in the ice.
These isotope ratios are sensitive to the temperature at the time they fell as snow from the clouds. The heavy isotope is lower in abundance during the colder winter snows than it is in snow falling in spring and summer. So the yearly layers of ice can be tracked by each of these five different indicators, similar to growth rings on trees.
A continuous count of layers exists back as far as 160,000 years. In addition to yearly layering, individual strong events (such as large-scale volcanic eruptions) can be observed and correlated between ice cores. A number of historical eruptions as far back as Vesuvius nearly 2,000 years ago serve as benchmarks with which to determine the accuracy of the yearly layers as far down as around 500 meters. As one goes further down in the ice core, the ice becomes more compacted than near the surface, and individual yearly layers are slightly more difficult to observe. For this reason, there is some uncertainty as one goes back towards 100,000 years. Ages of 40,000 years or less are estimated to be off by 2% at most. Ages of 60,000 years may be off by up to 10%, and the uncertainty rises to 20% for ages of 110,000 years based on direct counting of layers (D. Meese et al., J. Geophys. Res. 102, 26,411, 1997).
Recently, absolute ages have been determined to 75,000 years for at least one location using cosmogenic radionuclides chlorine-36 and beryllium-10 (G. Wagner et al., Earth Planet. Sci. Lett. 193, 515, 2001). These agree with the ice flow models and the yearly layer counts. Note that there is no indication anywhere that these ice caps were ever covered by a large body of water, as some people with young-Earth views would expect.