Latitudes at which ancient salt deposits occur show that Earth's magnetic field has always aligned along its rotation axis. One possible implication is that ancient global glaciations were not caused by a realignment of this axis.
In a paper of admirable scope and thoroughness that appears on page 51 of this issue1, David Evans analyses the magnetization locked into rocks associated with salts from all over the globe that have been deposited over the past 2,500 million years. Taking as a working model the 'geocentric axial dipole' — the idea that, averaged over thousands of years, the magnetic field at Earth's surface resembles the field of a magnet, or dipole, at Earth's centre2,3 — these magnetizations and this model provide clues to the past evolution and interplay of Earth's magnetism, climate and geography.
In the geocentric axial dipole model, the north and south poles of Earth's internal magnet are aligned along Earth's axis of rotation. This simple axial form is thought to be caused by rotational forces that guide the motions of Earth's conducting liquid core, and so constrain the average surface field. Under favourable circumstances, rocks become magnetized along the direction of the ambient geomagnetic field as they are formed. Thus, by sampling sequences of rocks with formation dates spanning several thousands of years, one can determine the past average direction of the field, the 'palaeolatitude' of the sampling locality and the position of the palaeomagnetic pole at the time. For the past 5 million years, these poles coincide with the present rotational pole; the giant dipole model has therefore been valid for at least this long.
For rocks of much earlier ages, the palaeomagnetic poles determined from rocks from different sampling sites are widely dispersed. This is the result of continental drift and seafloor spreading in the intervening period. If we restore the continents to their original positions using the geometrical methods of plate tectonics, palaeomagnetic pole positions agree very well4. Such corrections go back some 200 million years, and again imply that the geomagnetic field has been a geocentric dipole for that period.
But this evidence does not tell us that the field was also axial. To determine this, one first assumes that the geocentric axial dipole model holds, and determines the latitudes at which temperature-sensitive deposits were laid down from their magnetization directions, or, in the case of salts, those of similarly aged rocks. If these palaeolatitudes are compatible with the modern latitudes of similar deposits, the geocentric axial dipole model is likely to be valid.
The deposits that are the object of Evans's studies1 are known as evaporites. They comprise beds of, among other things, gypsum, anhydrite, halite and potassium salts, and are of huge economic importance. They were formed by intense evaporation of enclosed saline lakes or sea water. The conditions for their formation must therefore have been hot and dry, as expected typically in latitudes lower than 30°. Very near the Equator, however, it is too wet for them to form.
Evans shows that, consistently over the past 2,500 million years, evaporites have been deposited predominantly between latitudes 10° and 35° (Fig. 1, overleaf). This is a beautifully documented testament to uniformitarianism — the doctrine that today's geological processes have always occurred in a broadly similar manner.
Interest in the interplay between the geomagnetic field and ancient climate zones has been spurred by evidence in a wide range of latitudes, including at sea level near the Equator, that intermittent periods of glaciation covered the whole Earth between 750 million and 550 million years ago. Theories abound as to why this should have occurred, and one proposal5 is that Earth's obliquity might have changed drastically at the time. Obliquity is the tilt of Earth's equatorial plane to its ecliptic — the plane of its orbit around the Sun — and is currently 23.5°. If this tilt exceeds around 58°, high latitudes would get more solar heat than low latitudes, and this could account for low-latitude glaciations. The new data1 prove problematic for such models. Before and after the global glaciations, Evans consistently finds a strong low-latitude, off-Equator peak in evaporite deposition, indicating that obliquity then was not very different from now.
Evans's calculated palaeolatitudes do vary slightly through time, which he divides into four intervals running backwards. These are: interval 1, 250 million years ago (Myr) to the present; interval 2 (370–250 Myr); interval 3 (600–370 Myr); and interval 4 (2,500–600 Myr). The evaporite latitudes do tend to be lower in intervals 4, 3 and 2 than in interval 1, but the changes are irregular. There are two chief explanations for this variation. The first is that before interval 1 the geomagnetic field had long-term axial non-dipole components that endured unchanged for millions of years. These could have perturbed the geomagnetic field sufficiently to cause palaeomagnetic estimates of latitudes based on the geocentric axial dipole model to be systematically too low by as much as 10°. Alternatively, the field could have remained a simple axial dipole throughout, with the aberrations reflecting, for instance, continental drift or polar wandering.
Evans provides strong evidence for the validity of the geocentric axial dipole model throughout interval 1. Before that, he notes arguments6 favouring long-term axial non-dipole fields in interval 2, finds no grounds for them in interval 4, and seems undecided about interval 3.
I will focus on interval 2. Here, Evans accepts the previously advanced view6 that Alfred Wegener's grouping of continents into a supercontinent7, known as Pangaea A, persisted, not greatly changed, back through the entire interval. There are significant problems with such a long-lived Pangaea. Although there are firm correlations8 throughout interval 2 of very thick stratigraphic sequences within Gondwana and within Laurussia (respectively the clustered southern and northern continents that came together to form Pangaea A), there are no comparable correlations between Gondwana and Laurussia. Thus, placing these two supercontinents together in Pangaea A during interval 2 is problematic. Palaeomagnetic data9 for late interval 2 based on the assumption of an axial geocentric dipole field place portions of northern Gondwana at the same latitude as southern Laurussia, implying an impossible overlap of the two continents, one on top of the other, of around 1,000 km. But invoking the long-term axial non-dipole components would not remove this overlap9, because in such low latitudes their effect would be small and equal in both places. It is worth noting, too, that Evans's analysis does not explicitly require long-term axial non-dipole components in interval 2.
Evans makes no steady overall commitment to either of the contending explanations for the variations in the evaporite palaeolatitudes. I hope I will be forgiven for doing so. It seems reasonable to me to take the estimates of evaporite latitudes at face value and accept that, during the past 2,500 million years — apart from a relatively brief global glaciation at the end of interval 4 — hot and dry climates were typical of tropical, but not equatorial, latitudes, and that long-term non-dipole components have always been small or negligible. In other words, one should accept the geocentric axial dipole model as demonstrably successful for interval 1 and use it to settle questions of palaeogeography in earlier intervals. For example, displacing Gondwana to the east during interval 2 by around 3,500 km, a palaeogeography known as Pangaea B, reconciles the palaeomagnetic data with a feasible configuration of the early continents9.
Such palaeogeographic solutions are more generally testable than solutions involving long-term non-dipole fields: whereas the geomagnetic dynamo in early ages is an awfully remote item, ancient strata can be directly studied and the phenomena they reflect described. Most researchers accept interval 2 palaeogeography as settled. I disagree: on the principle of working from the known to the unknown, it could be just the issue that needs further critical consideration before we can confidently approach the question of earlier climate zones. Data of the extensive nature supplied by Evans1 are what is needed to inform such debates.