It is a fundamental tenet of geophysics that the Earth's magnetic field can exist in either of two polarity states: a "normal" state, in which north-seeking compass needles point to the geographic north, and a "reverse" state, in which they point to the geographic south. Geological evidence shows that periodically the field's polarity reverses, and that these reversals have been taking place at an increasing rate. Evidence also indicates that the field does not reverse instantaneously from one polarity state to another; rather, the process involves a transition period that typically spans a few thousand years. Though this much is known, the underlying causes of the reversal phenomenon are not well understood. It is generally accepted that the magnetic field itself is generated by the motion of free electrons in the outer core, a slowly churning mass of molten metal sandwiched between the Earth's mantle and its solid inner core. In some way that is not completely understood, gravity and the Earth's rotation, acting on temperature and density differences within the outer core fluid, provide the driving forces behind the generation of the field. The reversal phenomenon may be triggered when something disturbs the heat circulation pattern of the outer core fluid, and with it the magnetic field. Several explanations for this phenomenon have been proposed. One proposal, the "heat-transfer hypothesis," is that the triggering process is intimately related to the way the outer core vents its heat into the mantle. For example, such heat transfer could create hotter or cooler blobs of material from the inner and outer boundaries of the fluid core, thereby perturbing the main heat-circulation pattern. A more controversial alternative proposal is the asteroid-impact hypothesis. In this scenario an extended period of cold and darkness results from the impact of an asteroid large enough to send a great cloud of dust into the atmosphere. Following this climatic change, ocean temperatures drop and the polar ice caps grow, redistributing the Earth's seawater. This redistribution increases the rotational acceleration of the mantle, causing friction and turbulence near the outer core-mantle boundary and initiating reversal of the magnetic field. How well do these hypotheses account for such observations as the long-term increase in the frequency of reversal? In support of the asteroid-impact model, it had been argued that the gradual cooling of the average ocean temperature would enable progressively smaller asteroid impacts to cool the Earth's climate sufficiently to induce ice-cap growth and reversals. But theories that depend on extraterrestrial intervention seem less convincing than theories like the first, which account for the phenomenon solely by means of the thermodynamic state of the outer core and its effect on the mantle.