ools in a magma chamber within the crust below the stratovolcano. The processes that trigger the final eruption remain a question for further research. Possible mechanisms include: Magma differentiation, in which the lightest, most silica-rich magma and volatiles such as water, halogens, and sulfur dioxide accumulate in the uppermost part of the magma chamber. This can dramatically increase pressures. Fractional crystallization of the magma. When anhydrous minerals such as feldspar crystallize out of the magma, this concentrates volatiles in the remaining liquid, which can lead to a second boiling that causes a gas phase (carbon dioxide or water) to separate from the liquid magma and raise magma chamber pressures. Injection of fresh magma into the magma chamber, which mixes and heats the cooler magma already present. This could force volatiles out of solution and lower the density of the cooler magma, both of which increase pressure. There is considerable evidence for magma mixing just before many eruptions, including magnesium-rich olivine crystals in freshly erupted silicic lava that show no reaction rim. This is possible only if the lava erupted immediately after mixing since olivine rapidly reacts with silicic magma to form a rim of pyroxene. Progressive melting of the surrounding country rock. These internal triggers may be modified by external triggers such as sector collapse, earthquakes, or interactions with groundwater. Some of these triggers operate only under limited conditions. For example, sector collapse (where part of the flank of a volcano collapses in a massive landslide) can only trigger the eruption of a very shallow magma chamber. Magma differentiation and thermal expansion also are ineffective as triggers for eruptio