ANALYSING AIR BUBBLES IN ICE CORES
    Snow that is deposited on the surface of an ice sheet is gradually transformed into a mass of fine ice crystals, called firn. As more and more snow is deposited on top, the firn below is compressed to become ice; in the process, air bubbles are trapped in this ice. Air from above may still diffuse into the ice for a long time, but when the ice has become sufficiently compressed, and ice crystals have recrystallized to form coherent solid ice, air from above can no longer enter, and the trapped air remains shut off  from the atmosphere. The formation of air bubbles is described here.
    A crucial question is: how much younger is the trapped air than the surrounding ice?
    To answer this question, scientists have studied the distribution of air with differing mass. The nitrogen isotope N-15 is heavier than the normal N-14, and due to the effect of gravity, the lowest layers of air are enriched in N-15. But when the air bubbles are effectively shut off, this enriching is no longer possible. So when the ice is analysed, you find a gradient in the concentration of heavy air  downwards until a certain layer where the gradient levels off. This layer, then, marks the transition from loose firn to solid, impermeable ice. The age of that layer depends on the rate of snow accumulation at the top, and is therefore different in different ice sheets. The age difference between trapped air and the surrounding ice is typically about 500 years in some of the important ice cores, and the uncertainty on this age difference is probably about 20 years. Thus, in these cases, when the age of the air bubbles is corrected relative to the age of the surrounding ice, we end up with an uncertainty of 20 years on the age of the air bubbles. But where each annual layer of ice is thin, the age difference can be much larger, and the uncertainty on the age difference can be as much as 1,000 years.

Synchronization of ice cores
    If there is a sudden release of extra amounts of some gas into the atmosphere - for instance of CH4 (methane) - then the mixing of the atmosphere will cause the change in concentration to be equal all over the globe within about one year. That is, if we have a characteristic set of peaks in CH4 in air from an Antarctic ice core, and a similar set of peaks in ice cores from Greenland, then we know that these samples are simultaneous. This makes it possible to synchronize the time scales of Antarctic and Greenland ice cores with great accuracy (Blunier 1998).

Measurements of temperature
    Water containing oxygen-18 is heavier than water containing oxygen-16. This has the effect that the amount of this water that air can contain at a given temperature is low relative to the amount of ordinary water. When moist air is blown across an ice sheet, most of the heavier water has already condensed as precipitation, and that which falls as precipitation (snow) on the ice, is depleted in oxygen-18. The colder the air temperature when the water condenses, the greater is this effect. Therefore,     variations in the oxygen-18 content of ice is a measure of the air temperature when the ice formed, although there are some sources of errors which may modify the signal.
    The content of oxygen-18 in the O2 of the air does also vary with global temperature, but this is a more indirect effect (mainly due to differential consumption of heavier and lighter oxygen in the respiration of living organisms). Most of the precise "thermometers" of ancient temperatures are based on oxygen-18 in the ice, not in the air bubbles.
    Where very exact timing is crucial, it has also been possible to detect air temperature changes by measurements directly on the trapped air bubbles, studying the degree of spatial separation of air molecules with isotopes of different weight (nitrogen and argon isotopes) (Severinghaus et al. 1998).  Thereby  it is possible to relate changes in CO2 and CH4 directly to changes in temperature in the same air samples, especially because CH4 and N2 diffuse at the same rate into the ice. 

Measurements of carbon dioxide
    The concentration of carbon dioxide is measured directly in the air that is sampled from the air bubbles in the ice. This works well for Antarctic ice cores, but ice cores from Greenland contain much dust with impurities that may produce carbon dioxide after it is deposited; therefore, CO2 measurements in ice from Greenland show certain spurious results, and thus they are considered not very reliable.

Correlation between temperature, methane and carbon dioxide
    Recently analysed ice cores from Antarctica yield a long time series - going back 650,000 years - with data on temperature, CH4 and CO2. The variation over time of these parameters were shown for instance in Al Gore´s movie, An Inconvenient Truth, and are also depicted in the British version of Lomborg´s book "Cool it!". The curves show an extraordinarily close correlation between variations in all three parameters. Every time temperature goes up, CH4 and CO2 go up as well. However, upon very close inspection it is seen that the temperature signal is ahead of the other two signals by a time difference which is typically about 1,000 years. The most exact estimate is a lag of 800 years ± 200 years (Caillon et al., 2003). The fact that the CO2 signal lags behind the temperature signal is interpreted by climate change skeptics to say that temperature drives CO2, not the other way around. This conclusion is, however, so simplified that it is wrong.
    What is forgotten is that this phenomenon is seen only in ice cores from Antarctica, not in ice cores from Greenland. When the results for the last 50,000 years are synchronized between Greenland and Antarctica, it turns out that each warming event typically starts slowly in Antarctica, and then leads to at more sudden warming in Greenland 1,000 - 2,500 years later (Blunier 1998). This means that when the warming trend finally reaches Greenland, CO2 concentrations are already rising. In other words, in the Northern Hemisphere the CO2 signal does not lag behind the temperature signal. This was even more clear when the last glaciation terminated. The temperature rise on Antarctica started 20,000 -19,000 BP (Blunier 1998). CO2 concentrations in the Greenland ice (and in Antarctica) started to rise about 18,000 BP (Anklin et al., J. geophys. res. 102 (C12): 26,539-26,545). The temperature rise on the Northern Hemisphere started very abruptly at the end of the oldest Dryas at about 14,700 years ago, and the temperature rise in the tropics seems to have happened 20-30 years later than in the north (Severinghaus & Brook 1999). Thus, CO2 had been on the rise for about 3,300 years before the warming reached the Northern Hemisphere and the tropics. A very similar time course happened at the termination of another ice age about 245,000 years ago, when CO2 had been on the rise for about 4,200 years before the warming reached the Northern Hemisphere (Caillon et al., 2003).

The influence of changes in the Earth´s orbit
   It is generally believed that the starting and ending of ice ages is triggered by cyclical changes in the Earth´s orbit.  However, these changes contribute to climate changes with a forcing which is calculated to be only around 0.7 watt/m² (link). This is too little to cause the full temperature jump (4-5° C) which separates interglacial periods from glacials, but it is enough to initiate a chain reaction as explained here: The warming at the start of an interglacial is caused by orbital changes which gives slightly more insolation to the seas of the southern hemisphere. This means some warming at Antarctica - which is recorded in the ice cores - and warming of the large masses of sea water on the southern hemisphere. Somehow this warming leads to a rise in atmospheric CO2, partially because when the water gets warmer, some of the CO2 is released into the atmosphere.This CO2, in turn, augments the greenhouse effect and amplifies the temperature rise that is already underway. It has been calculated that out of the total temperature rise at the end of an ice age, between 40 and 65 % is explained by changes in greenhouse gases (CO2 and CH4) (link).Without changes in greenhouse gases, the full switch from glacial to interglacial would not be possible.

The transfer of warming from the southern to the northern hemisphere.
    The time lag between the temperature rise in the Antarctic, and the rise in atmospheric CO2 is, as stated, about 800 years. This is approximately the same time that it takes for the oceans to exchange surface water and deep sea water. It is therefore believed that a chain of events take place once the deep sea water has become warmer, with concomitant changes in ocean currents (Raynaud et al. 2000). Because the "great ocean conveyor" transports deep see water across the equator, heat will be transferred to the Northern Hemisphere, once it has reached the deep sea. The northward movement of deep sea water may have happened primarily in the Pacific Ocean, like it is today. This has then secondarily affected the thermohaline circulation in the North Atlantic, which pulls warmer water from the tropics towards northern latitudes. The heat energy necessary for this circulation has probably not been available during the glacial period, but once heat is transferred across the equator, and once CO2 in the atmosphere has started to rise, the circulation could start. Once it started, heat was led towards Greenland, which warmed. The melting of ice on the Northern Hemisphere decreased the albedo and led to further acceleration of the warming there. Simultaneously with this chain of events, there was a slight cooling in the southern Hemisphere, which further corroborates that natural temperature changes at the two poles are not precisely in phase.
    What happens now, however, is different from anything that happened in the past: atmospheric CO2 is increased ahead of natural warming processes, and CO2 drives the temperature rise simultaneously all over the globe.