The concept of the North Magnetic Pole arose from the desire of early European navigators to explain the directional properties of the compass. The compass was in use in China at least as early as the 1st century and appears to have been imported into Europe by the 12th century; the earliest European reference dates from 1190. However, whereas the Chinese considered the compass a south-pointing device, Europeans considered it north-pointing. This change in orientation would prove important in the development of theories about the nature of the Magnetic Pole.

The first detailed descriptions of both a floating and a pivoted compass appeared in the Epistola of Petrus Perigrinus written in 1269. In this remarkable work Perigrinus did more than just describe the construction of a compass. He described experiments performed on a sphere of lodestone, predating the more famous work of William Gilbert by more than 300 years. Peregrinus made three important discoveries:

1. the dipolar nature of the magnet
2. the fact that the magnetic force is vertical at the poles
3. the fact that the force is strongest at the poles

Peregrinus considered that the ability of magnetic needle to point north was not a fundamental property of the magnet or of the earth. Instead, he resorted to one of two theories prevalent at the time. Since the compass needle points toward the pole star, which is situated on the celestial axis about which the 10 heavenly spheres rotate, it must receive its properties from that same star. The second theory, which was to become widespread, is that the compass needle is attracted to a magnetic lodestone mountain located at the north pole.

It was generally assumed that the magnetic mountain was located at the geographic pole, so the discovery of magnetic declination, that the compass does not point true north, posed a problem which was solved by placing the magnetic mountain some distance from the geographic pole. The great map maker, Gerhard Mercator, attempted to locate the magnetic pole more precisely by determining the intersection point of great circles derived from values of magnetic declination obtained at different locations. Mercator first tried this in 1546, but soon found that his great circles did not all intersect at a single point. He solved this problem by invoking two magnetic poles. This accompanying figure shows the Arctic insert on Mercator's famous map of 1569 which clearly shows two magnetic poles, separated by 500 km.

What Mercator had unknowingly discovered was the non-dipole nature of the earth's magnetic field, which means that magnetic meridians do not follow great circles between given points and the magnetic pole.

We owe our present definition of the magnetic poles to William Gilbert, who wrote his famous book de Magnete in 1600. Gilbert owed a lot to Petrus Peregrinus and to Robert Norman, who had discovered inclination in 1576, but Gilbert's own contributions to the science of magnetism were considerable. He was the first to postulate that the earth itself was a great magnet, and the force that directs a compass needle lies within the earth, not in some lodestone mountain or in the celestial spheres. He defined the magnetic poles as the two places on the surface of the earth where a magnetized needle would stand vertically.

Gilbert believed that the magnetic poles and geographic poles were coincident; his a priori cosmological beliefs coupled with his experiments on the lodestone terrella dictated this. He certainly knew about declination but said it was due to a greater amount of magnetic material in the continents relative to the oceans, which caused the compass to decline towards the continents. In modern language, declination was caused by large-scale crustal anomalies. The data at the time did not contradict this, but eventually it was shown that this part of Gilbert's theory was wrong.

As the number of magnetic declination observations increased, the concept of a magnetic pole or a single pair of poles came under criticism because the theoretical pattern of magnetic declination, essentially that of a dipole, did not match the observed declination. The mathematician Leonhard Euler tried offsetting the magnetic axis from the centre of the Earth so that the two magnetic poles were not diametrically opposite, but that was not enough. What seemed to be required were more poles.

The great British astronomer Edmund Halley, who also published the first magnetic declination chart of the Atlantic Ocean in 1702, postulated that there were two north magnetic poles and two south magnetic poles. Two of these poles he placed on the surface of the Earth; the other two he placed on an inner sphere about 800 km below the surface. This arrangement allowed him to explain the observed pattern of declination and, by allowing the inner and outer spheres to rotate at slightly different speeds, he could explain the secular variation of declination.

In the early part of the nineteenth century the four-pole theory was promoted by Christopher Hansteen. Hansteen had many more declination observations than Halley, as well as inclination and total force observations. In particular, he noted that the shape of the magnetic equator could not be explained by a single magnetic axis, nor could the lack of correlation between total force and inclination. Hansteen believed that the observed pattern of the magnetic field could be explained by two magnetic axes, each with two poles, but his definition of what constituted a magnetic pole differs from the definition we use today. According to Hansteen, a magnetic pole is not a point where the magnetic field is vertical, nor is it a point towards which a compass needle points. Rather, it is a "point of force", a place where magnetic intensity is a maximum.

In fact, in the early nineteenth century there were three different definitions attached to the term "magnetic pole". Some interpreted it to mean the point at which magnetic meridians converged; others, including Hansteen took it to mean the area at which magnetic intensity was a maximum; and a third group, which included James Ross, defined it to mean the point of vertical dip. Were the earth's magnetic field perfectly dipolar, all three definitions would correspond to the same point, but by the early 19th century the complexity of the magnetic field was well appreciated, including the fact that there were two areas of maximum intensity in the northern hemisphere. In modern terms, by the early 19th century it was well known that the magnetic field was too complex to be explained by a single dipole, and researchers such as Hansteen were formulating theories to account for the non-dipole part. This they chose to do by adding a second dipole. Using multiple dipoles to model the magnetic field is perfectly valid procedure, but analyses carried out in the 1960s showed that up to 35 radial dipoles are necessary to model the field with acceptable accuracy.

In 1839 Frederick Gauss developed the method of spherical harmonic analysis for describing the magnetic field. Magnetic poles were not required, nor did they play any part in the analysis. The existence of two magnetic poles, one in each hemisphere, was a by-product of the analysis, but the definition of a magnetic pole was restricted to mean the region on the Earth's surface in which the horizontal intensity is zero and inclination is ±90°. Gauss made it clear that the concept of a magnetic axis joining the two poles has no basis in fact.

Not everyone agreed with Gauss's idea at the time, but today his method of spherical harmonic analysis is universal, as is his concept of the magnetic poles.