In the early 20th century geologists first noticed that some volcanic rocks were magnetized in a direction opposite to what was expected. The first examination of the timing of magnetic reversals was done by Motonori Matuyama in the 1920s, who observed that there were rocks in Japan whose magnetic fields were reversed and those were all of early Pleistocene age or older. At the time he published his proposal suggesting that the magnetic field had been reversed, the magnetic field itself was poorly understood so there was little interest in the possibility that it had reversed.
Three decades later, theories existed of the cause of the magnetic field and some of these included the ability for the field to be reversed. Most paleomagnetic research in the late 1950s was examining the wandering of the poles and continental drift. Although it was discovered that some rocks would reverse their magnetic field while cooling, it became apparent that most magnetized volcanic rocks contained traces of the Earth's magnetic field at the time the rock cooled. At first it seemed that reversals happen every one million years, but during the 1960s it became apparent that the time between reversals is erratic.
During the 1950s and 1960s research ships gathered information about variations in the Earth's magnetic field. Because of the complex routes of cruises, associating navigational data with magnetometer readings was difficult. But when data was plotted on a map, it became apparent that there were remarkably regular and continuous magnetic stripes across the ocean floors.
In 1963 Frederick Vine and Drummond Matthews provided a simple explanation, by combining the seafloor spreading theory of Harry Hess with the known time scale of reversals: if new sea floor acquired the present magnetic field, spreading from a central ridge would produce magnetic stripes parallel to the ridge. Canadian L. W. Morley independently proposed a similar explanation in January 1963, but his work was rejected by the scientific journals Nature and Journal of Geophysical Research, and not published until 1967 in the literary magazine Saturday Review.
Starting in 1966, Lamont-Doherty Geological Observatory scientists found the magnetic profiles across the Pacific-Antarctic Ridge were symmetrical and matched the pattern in the north Atlantic's Reykjanes ridges. The same magnetic anomalies were found over most of the world's oceans, and allowed estimation of the timing of the creation of most of the oceanic crust.
Through analysis of palaeomagnetic data, we now know that the field has reversed its orientation tens of thousands of times since its formation very early on in earth history. With the increasingly accurate Global Polarity Timescale (GPTS) it has become apparent that the rate at which reversals occur has varied considerably throughout the past. During some periods of geologic time (e.g. Cretaceous Long Normal), the Earth's magnetic field is observed to maintain a single orientation for tens of millions of years. Other events seem to have occurred very rapidly, with two reversals in a span of 50 thousand years. The last reversal was the Brunhes-Matuyama reversal approximately 780 thousand years ago.
In some simulations, this leads to an instability in which the magnetic field spontaneously flips over into the opposite orientation. This scenario is supported by observations of the solar magnetic field, which undergoes spontaneous reversals every 7-15 years (see: solar cycle). However, with the sun it is observed that the solar magnetic intensity greatly increases during a reversal, whereas all reversals on Earth seem to occur during periods of low field strength.
Present computational methods have used very strong simplifications in order to produce models that run to acceptable time scales for research programs.
A minority opinion, held by such figures as Richard A. Muller, is that geomagnetic reversals are not spontaneous processes but rather triggered by external events which directly disrupt the flow in the Earth's core. Such processes may include the arrival of continental slabs carried down into the mantle by the action of plate tectonics at subduction zones, the initiation of new mantle plumes from the core-mantle boundary, and possibly mantle-core shear forces resulting from very large impact events. Supporters of this theory hold that any of these events could lead to a large scale disruption of the dynamo, effectively turning off the geomagnetic field. Because the magnetic field is stable in either the present North-South orientation or a reversed orientation, they propose that when the field recovers from such a disruption it spontaneously chooses one or the other state, such that a recovery is seen as a reversal in about half of all cases. Brief disruptions which do not result in reversal are also known and are called geomagnetic excursions.
Because the magnetic field is present globally, finding similar patterns of magnetic variations at different sites is one method used to correlate age across different locations. In the past four decades great amounts of paleomagnetic data have been accumulated about current seafloor ages (up to ~250 Ma) to such an extent that such data have become an important and convenient tool used to estimate the age of geologic sections in the field. It is, however, not an independent dating method, but is dependent on "absolute" age dating methods like radioisotopic systems to derive numeric ages. It has become especially useful to metamorphic and igneous geologists where the use of index fossils to estimate ages is seldom available.
An interesting trend can be seen when looking at the frequency of magnetic reversals approaching and following the Cretaceous Long Normal. The frequency steadily decreased prior to the period, reaching its low point (no reversals) during the period. Following the Cretaceous Superchron the frequency of reversals slowly increased over the next 80 million years, to the present.
At present, the overall geomagnetic field is becoming weaker at a rate which would, if it continues, cause the dipole field to temporarily collapse by 3000–4000 AD. The South Atlantic Anomaly is believed by some to be a product of this. The present strong deterioration corresponds to a 10–15% decline over the last 150 years and has accelerated in the past several years; however, geomagnetic intensity has declined almost continuously from a maximum 35% above the modern value achieved approximately 2000 years ago. The rate of decrease and the current strength are within the normal range of variation, as shown by the record of past magnetic fields recorded in rocks.
The nature of Earth's magnetic field is one of heteroscedastic fluctuation. An instantaneous measurement of it, or several measurements of it across the span of decades or centuries, is not sufficient to extrapolate an overall trend in the field strength. It has gone up and down in the past with no apparent rhyme or reason. Also, noting the local intensity of the dipole field (or its fluctuation) is insufficient to characterize Earth's magnetic field as a whole, as it is not strictly a dipole field. The dipole component of Earth's field can diminish even while the total magnetic field remains the same or increases.
The Earth's magnetic north pole is drifting from northern Canada towards Siberia with a presently accelerating rate — 10km per year at the beginning of the 20th century, up to 40km per year in 2003. It is also unknown if this drift will continue to accelerate.
Glatzmaier and collaborator Paul Roberts of UCLA have made a numerical model of the electromagnetic, fluid dynamical processes of Earth's interior, and computed it on a Cray supercomputer. The results reproduced key features of the magnetic field over more than 40,000 years of simulated time. Additionally, the computer-generated field reversed itself.
Because the magnetic field has never been observed to reverse by humans with instrumentation, and the mechanism of field generation is not well understood, it is difficult to say what the characteristics of the magnetic field might be leading up to such a reversal. Some speculate that a greatly diminished magnetic field during a reversal period will expose the surface of the earth to a substantial and potentially damaging increase in cosmic radiation. However, Homo erectus and their ancestors certainly survived many previous reversals. There is no uncontested evidence that a magnetic field reversal has ever caused any biological extinctions. A possible explanation is that the solar wind may induce a sufficient magnetic field in the Earth's ionosphere to shield the surface from energetic particles even in the absence of the Earth's normal magnetic field .
Although the inspection of past reversals does not indicate biological extinctions, present society with its reliance on electricity and electromagnetic effects (e.g. radio, satellite communications) may be vulnerable to technological disruptions in the event of a full field reversal.