Reading The Magnetic Record From Rocks

The short answer is that rocks record the direction and strength of the Earth's magnetic field at the moment they form, and we have been reading those records for decades. When molten rock cools through its Curie temperature, tiny magnetic minerals inside it align with the ambient field and lock into place. That remanent magnetism stays put unless the rock is heated again or physically disturbed. Sediments do something similar as they settle on the ocean floor. This is what lets us reconstruct past field behavior going back hundreds of millions of years. The detailed mechanism involves measuring samples in a clean lab environment, demagnetizing them step by step, and extracting the characteristic remanent magnetization that survives the process. I used to run paleomagnetic measurements myself, and the work is straightforward if you accept that it takes time. A single sample line through an AF demagnetizer and spinner magnetometer runs about twenty minutes. You need a couple dozen samples per site to get a reliable direction. Sites are collected across geographic and stratigraphic spread so you are not accidentally reading overprinted or locally magnetized material.

Scientists Know The History Of Earths Magnetic Field Because Of Paleomagnetic Records In Igneous And Sedimentary Rocks

The records come from several sources. Lava flows preserve the field direction at the time each flow solidifies. Sediment cores give continuous sections where individual layers date precisely to the depositional age. Thermoremanent magnetization in volcanic rocks is generally more stable than chemical remanent magnetization in sediments, which means igneous sequences tend to carry cleaner signals. That said, both record enough to build a usable composite history. One thing beginners consistently miss is that a single rock sample does not represent the field at one moment. It represents an average over whatever period the rock was cooling or depositing. Basaltic flows can cool in weeks, which gives good temporal resolution. Thick lava plateaus or slow-cooled plutons can smear the signal over thousands of years. Sediments vary even more depending on accumulation rate. A core site depositing a millimeter per century will blur rapid field changes far more than a site accumulating a centimeter per century. This matters when you try to resolve individual geomagnetic excursions. Another counter-intuitive detail is that the field does not just flip. Between full reversals there are excursions, non-dipole events, and gradual drift in the poles. The geomagnetic polarity timescale is built by stacking marine magnetic anomalies along with dated polarity intervals from land sections. Marine anomalies are what first made this feasible on a global scale. As seafloor spreads, newly formed crust records the field in striped patterns of positive and negative anomaly. Those stripes correlate across oceans and tie directly to the dated polarity timescale. The practical side of doing this work involves a few specific choices that matter more than people expect. The first is demagnetization protocol. You typically run alternating field steps starting at five to ten milliteslas and going up to forty or fifty milliteslas depending on sample coercivity. Some samples need thermal demagnetization up to 600 or 700 degrees Celsius. You plot the vector end-points on an orthogonal diagram and look for a straight line toward the origin. That line is your characteristic direction. Anything that fans out or curves is either a secondary overprint or multiple magnetization components superimposed on each other. I encountered a specific problem with a set of basalt samples from a Late Cretaceous site that looked clean at first glance. The natural remanent magnetization directions clustered tightly, which is usually a good sign, but the intensity values varied by a factor of six across samples from the same flow. That should have been the red flag. High variation like that often points to post-emplacement heating or impact metamorphism locally resetting part of the magnetization. I cut a subset of the samples for thin-section analysis and found fine-grained alteration halos around magnetite grains. The original thermoremanent magnetization was still there in the unaltered cores, but the bulk measurement was biased. The workaround was to use stepwise demagnetization combined with thermal treatment at low temperature increments rather than pushing directly into high alternating fields. The altered component came off at lower demagnetization steps, leaving the higher-coercivity primary remanence intact. The resulting directions shifted by about twelve degrees compared with the naive average, which changed the apparent paleolatitude of the site by roughly three degrees. That is not a trivial difference when you are tying a polarity interval to a specific formation age. What the data actually show is that the field has reversed many times. The most recent reversal was the Brunhes-Matuyama event about 780,000 years ago. Before that, the Matuyama chron contained dozens of short normal-polarity subchrons. Going further back, the Cretaceous Normal Superchron lasted roughly thirty million years with no reversals at all. Reversal frequency is not constant. There are periods of high activity and long quiet intervals. The mechanism driving this is the geodynamo in the liquid outer core, which is chaotic by nature. We can model it with numerical simulations, but those models still struggle to reproduce reversal statistics that match the geological record without tweaking parameters. There are real limitations to this approach. Older rocks are harder to work with because they have experienced more metamorphic events, tectonic strain, and alteration. Metamorphic overprints can completely erase the original signal or replace it with a new one that records only the field during the metamorphic event. In my experience, about one in four sites from Precambrian sequences turns out to be unrecoverable after demagnetization attempts. You just move on to the next site. Another limitation is that the resolution drops significantly for time intervals with slow sedimentation or sparse volcanic activity. If you need to resolve something happening over a few hundred years, you are usually stuck unless you have a very favorable sedimentary section or a volcanic sequence with rapid eruptions. Most of the polarity timescale is calibrated to tens of kiloyears at best, and often worse. People sometimes assume that because the field weakens before a reversal, we should see a clear warning signal in the present-day data. The current decay of the dipole moment is real, but it is within the range of normal variability. The field has weakened and recovered many times without reversing. Interpreting today's decline as an imminent reversal signal is not supported by the paleomagnetic record. If you are trying to build your own polarity sequence or test a site, the standard workflow is collecting oriented samples with a drone compass and inclinometer, cutting them into standard cylinders, measuring the NRM, running demagnetization steps, analyzing the vector components, and then placing the resulting mean direction into a regional or global reference frame. The reference frames change periodically as more data become available. The most recent IGRF and its paleomagnetic counterparts incorporate results from thousands of sites, which is why published ages and pole positions shift slightly between releases. The bottom line is that the history of the Earth's magnetic field comes from measuring remanent magnetization in rocks and sediments, cleaning those signals through demagnetization, and stacking the results into a timeline anchored by radiometric dates and marine magnetic anomalies. It works well enough to give us a coherent picture spanning hundreds of millions of years, and it has clear limits where the signal is overwritten or too coarse to resolve short events.