One we have dated a sufficient number of rocks and measured the orientation of the magnetism they contain, we can build up a picture of how the position or apparent position of the poles over time.
So if we are then faced with a rock the date of which we do not know, then we do know (of course) the latitude and longitude at which we found it, and we can measure the orientation of its magnetism, and so we can look at the global picture we've built up of continental drift, and so figure out when the rock must have formed in order to have its magnetism oriented in just that direction.
The rocks that yield these anomalously low readings therefore must have formed at a time when the Earth's magnetic field was reversed—oriented in such a way that the north magnetic pole was roughly where today's south magnetic pole is, and vice versa.
These magnetic reversals, in which the direction of the field is flipped, are believed to occur when small, complex fluctuations of magnetic fields in the Earth's outer liquid core interfere with the Earth's main dipolar magnetic field to the point where they overwhelm it, causing it to reverse.
As noted in a previous article, magnetic reversals come at irregular intervals.
The models show a ridge (a) about 5 million years ago (b) about 2 to 3 million years ago and (c) in the present.
Paleomagnetism (or palaeomagnetism in the United Kingdom) is the study of the record of the Earth's magnetic field in rocks, sediment, or archeological materials.
This means that the pattern of normal and reverse polarity in an assemblage of rocks can be distinctive in the same way (though for a completely different reason) that growth rings in a tree can be distinctive.
We might, for example, see a long period of reverse polarity, followed by six very quick switches of polarity, followed by a long period of normal polarity; and this might be the only time that such a thing occurs in our timeline.