Sunteți pe pagina 1din 10

SPECIAL FEATURE: ARCHAEOLOGY

www.iop.org/journals/physed

Archaeomagnetic dating
Paul Linford
English Heritage, Centre for Archaeology, Fort Cumberland, Fort Cumberland Road, Eastney, Portsmouth PO4 9LD, UK E-mail: paul.linford@english-heritage.org.uk

Abstract Some naturally occurring minerals possess a permanent magnetization. Certain processes such as sedimentation or kiln-ring can cause the particles in structures made of such materials to align themselves with the direction of Earths magnetic eld at the time. This direction has varied over the last few thousand years in ways that can be traced from known records, so it provides a means of dating such structures if they have remained in their original orientationas a kiln would, for example.

Introduction
Archaeomagnetic dating is less frequently used than other physical dating techniques, such as carbon-14 dating, owing to limitations on the materials to which it can be applied. Nevertheless, it can provide valuable, often precise, chronological evidence for archaeological features, particularly where no suitable remains are present to be dated by other means. It depends upon two important physical phenomena: (i) The Earth spontaneously generates a magnetic eld, which uctuates in intensity and direction with time. (ii) Under certain conditions naturally occurring magnetic minerals can become permanently magnetized according to the magnetic eld pertaining at that time.

The Earths magnetic eld


The details of the mechanism by which the Earths magnetic eld is generated are not completely understood. It appears to be associated with a region 3000 km beneath the planets surface in the outer core, which is composed of slowly churning molten iron and nickel. This layer is trapped between the solid inner core at the centre of the planet and the mantle, another
0031-9120/04/020145+10$30.00

solid layer extending from 3000 km to about 40 km beneath the surface. It is now generally accepted that free electron circulation within the convecting outer core creates the magnetic eld, which behaves as a self-sustaining dynamo. The uid motions that drive this dynamo derive from the Earths rotation along with gravitational and thermodynamic effects in and around the core. This results in an approximately dipolar magnetic eld at the Earths surface, as if a large bar magnet was situated at its centre with its long axis aligned almost parallel with the Earths rotational axis (gure 1). Near the equator the direction of the eld lines is horizontal (parallel to the surface of the Earth); however, as one of the two magnetic poles is approached they rise out of or dip into the ground at an increasingly steep angle. This angle is known as the angle of dip or inclination. The positions of the Earths magnetic poles do not exactly coincide with its geographic poles, and at present the axis of the dipolar eld is inclined at 11.5 to the Earths rotational axis. Because of this, the direction indicated by a compass needle at an arbitrary point on the Earths surface will generally deviate from the direction of geographic, or true, north. The angle in the horizontal plane between magnetic north and true north is known as the magnetic
PHYSICS EDUCATION 39 (2)

2004 IOP Publishing Ltd

145

P Linford

magnetic north pole angle of dip, I

geographic north pole

S N

eddy current

currents generating main dipole field mantle fluid outer core

Figure 1. The Earths main dipolar magnetic field is depicted with dashed lines. This is generated by electric current circulation in the outer core (shown in red) and is similar to the field that would be produced by a bar magnet located at the Earths centre tilted off-vertical by about 11.5. Eddy currents near the core/mantle boundary perturb this main field. The angle of dip (or inclination) is the angle that the field lines make with the horizontal plane where they cut the Earths surface.

declination. Over geologic timescales the position of the magnetic poles appears to precess about the geographic poles. This movement is caused by the same forces that generate the Earths eld. Superimposed upon the generally circulatory movement of the magnetic poles is an apparently random element known as secular variation. This is believed to be due to eddy currents in the Earths uid core and to the movement of charged particles in the upper atmosphere. As a result, the magnetic pole position calculated from measurements of declination and inclination at one point on the Earths surface will not match exactly that calculated from measurements made at another position. For archaeomagnetic dating, this has the consequence of requiring the compilation of separate calibration curves of the variation of the Earths eld with time for different regions, each about 1000 km in diameter.
146
PHYSICS EDUCATION

As well as changes in the position of the magnetic poles, the intensity of the Earths eld varies with time. Evidence from magnetizations recorded in igneous rocks indicates that periodically, every few hundred thousand years, the eld intensity decreases to nothing then increases again with the polarity reversed (i.e. the north and south poles change places) (Hoffman 1988). Again, a random component of variation is superposed on this trend over shorter timescales. The eld intensity determines the strength of the attraction of a compass needle to the magnetic poles and the strength of magnetization acquired by magnetic minerals.

Remanent magnetism
Some naturally occurring minerals are ferrimagnetic, possessing a permanent or remanent magnetization. By far the most prevalent of these are
March 2004

Archaeomagnetic dating

Figure 2. Thermoremanent magnetization. Initially magnetic domains within a sample are magnetized in random directions that cancel out (top picture). As the sample is heated the domains demagnetize as the temperature exceeds their blocking temperatures (second and third pictures down). On cooling, the domains remagnetize in a direction close to the prevailing ambient magnetic field, resulting in a net magnetization within the sample (bottom two pictures).
March 2004

the iron oxides, haematite, maghaemite and magnetite, which occur in most soils, clays and as trace components in many types of rock (Thompson and Oldeld 1986). In crystals of these minerals, quantum mechanical exchange interactions between neighbouring atoms force all the unpaired electron magnetic moments to align, resulting in a net spontaneous magnetization (Tauxe 2002). Such alignment will occur over a region of the crystal known as a magnetic domain and the shape and size of these domains will depend on the structure and size of the crystal as well as the impurities within it. Each may correspond to one physical grain of the mineral crystal (as occurs in haematite), or a single grain may be divided into several magnetic domains (as can occur in magnetite). Magnetization within each domain will be in one of the two directions parallel to its easy axis, a direction determined by domain shape and the underlying crystal structure. As each domain will typically be magnetized in a different, randomly orientated, direction, a macroscopic sample of the mineral containing a large number of domains will usually exhibit no net magnetization. However, if the mineral is heated, thermal agitation of the crystal structure leads to a diminution of the spontaneous magnetization in each domain until, at a certain critical temperature, known as the blocking temperature, it disappears entirely (gure 2). On cooling, each domain will remagnetize in one of the two directions parallel to its easy axis: that adopted will be the one closest to any ambient magnetic eld direction. Although most individual domains magnetizations will not be exactly aligned with the ambient eld direction, they will tend to favour it on average. Thus, after heating, the mineral will exhibit a net thermoremanent magnetization in the direction of the prevailing magnetic eld at the time it cooled. The blocking temperatures of different magnetic domains vary. Maximum blocking temperature is limited by the Curie temperature of the particular mineral involved (585 C for magnetite and 675 C for haematite) but considerations of grain size and crystal structure can reduce it below this limit. Indeed even without heating, and in the absence of an external magnetic eld, some domains will spontaneously reverse their directions over time. In the idealized case the probability, p , that this will happen varies as
PHYSICS EDUCATION

147

P Linford

(Tauxe 2002, N eel 1955) p = et/ where = 1 Kv/kT . e C

Here C is a frequency factor around 1010 s1 , K is an anisotropy constant, v is the volume of the magnetic domain, k is Boltzmanns constant and T is the ambient temperature. The blocking temperature is dened as that temperature which results in a value of between 102 and 103 seconds. Naturally occurring samples of rocks and clays will contain a heterogeneous mineral composition and thus a spectrum of blocking temperatures. To be capable of retaining a stable thermoremanent magnetization, they must contain a high proportion of magnetic domains with blocking temperatures above 200300 C. In addidion to thermoremanent magnetization, a second mechanism, called depositional remanent magnetization, also occurs in archaeological contexts. This involves water-borne sediment particles that possess a weak overall magnetization, often because they are composed of thermoremanently magnetized minerals (although chemical effects during crystal growth can also result in remanent magnetization). If suspended in still water, they will attempt to rotate so that their directions of magnetization align with the prevailing magnetic eld (gure 3). As gravitational forces pull the particles to the bed of the body of water they will form a layer magnetized in the direction of this ambient eld. As more sediment accumulates above this layer, frictional forces caused by its weight lock the magnetized particles in place so that they are no longer free to rotate and realign themselves. This situation can occur when alluvial sediments accumulate on relatively still lakebeds and a stratigraphic sequence of magnetic layers results, recording changes in the Earths magnetic eld over time.

Archaeomagnetic dating techniques


If the past history of the changes in the Earths magnetic eld is known, there are two principal ways to date an archaeological artefact, structure or deposit that has acquired a remanent magnetization at some time in the past.
148
PHYSICS EDUCATION

Figure 3. Depositional remanent magnetization. Sediment particles, each with a weak magnetization, settle out of still water. As they fall through the water column they rotate to align their internal magnetization directions with the Earths field (upper three pictures). Once settled on the bed of the body of water, the weight of sediment accumulating on top of the particles locks them in place, leaving a layer magnetized in the direction of the Earths field.

The rst method is to infer from the strength of the magnetization in the artefact, the intensity
March 2004

Archaeomagnetic dating

0 -15

-160

-170

180

170

160

15

4 -1

14

0 -15
0 13

-160

-170

180

170

160

15

4 -1

14

0 13 0

-1 30

-1 30
12

12

-12

-12

-11 0

11 0

11 0

-110

100

10 0

-100

-100

80

70

60

50

-4

0
-30

40

-4

0
-30

40

-20

30
-10
0 oE

10

20

-20

30
-10
0 oE

10

20

Figure 4. Movement of the apparent position of the Earths magnetic north pole with time, based upon archaeomagnetic measurements made in the UK. The picture on the left shows movement from the Bronze Age until 500 AD, the right-hand picture shows movement since then.

of the Earths eld at the time that it acquired its remanence. As the direction of the eld is not involved, this has the attraction that it allows portable archaeological objects such as potsherds to be dated. A technique for doing this was developed by Thellier and Thellier (1959). This involves repeatedly heating the sample in a zero eld to a number of increasing temperature stages and measuring the intensity of the magnetization remaining after each stage. Once all the magnetization is removed, the process is repeated but with the sample exposed to a known reference magnetic eld during each heating. From a comparison of the results it is possible to infer the strength of the Earths eld at the time the sample was originally heated in antiquity. However, many measurements are required for this technique, as compared with directional dating, and more sources of error are involved. These can arise from: sample inhomogeneity, sample anisotropy, differences between the original and laboratory ring atmosphere and differences between the original and laboratory heating and cooling rates (Tarling 1983). Given these problems, most archaeomagnetic research in the UK has been directed towards the development of a technique involving the direction of the Earths magnetic eld. This involves establishing the magnetic north pole position indicated by the declination and inclination of
March 2004

the eld within the sample and determining when in the past magnetic north was in that position. Clearly, the artefact must have remained in exactly the same position as it was when it acquired its remanence, limiting the types of object that can be dated to non-portable structures. To this end, analysis of lake sediment data (depositional remanence), a large number of well dated archaeological structures (thermoremanence), as well as direct compass measurements from the last 400 years, has led to the construction of the United Kingdom archaeomagnetic calibration curve (Clark et al 1988). Based upon these data, gure 4 shows the variation in the apparent position of the magnetic north pole as viewed from the United Kingdom over the past 3000 years. Thus, in principle, the date at which an unmoved archaeological object acquired its remanence can be inferred by measuring the declination and inclination of its magnetization, determining the corresponding magnetic pole position and comparing this with the dated timeline.

What can be dated?


Directional archaeomagnetic dating imposes three constraints on the types of archaeological features that can be dated. They must
PHYSICS EDUCATION

50

60
-5 0

70

80
-60

-90
-80

-90

90

90

-8 0

-70
-60

-70

-5 0

149

P Linford

Figure 5. Base of a Roman pottery kiln (1.5 m in diameter) constructed of red clay, discovered at Heybridge in Essex and dated to the second century AD.

(i) contain magnetic minerals capable of carrying a stable remanent magnetization, (ii) have experienced a remanence-inducing event at some time in their history (e.g. heating above a blocking temperature or non-turbulent sediment deposition), (iii) have remained undisturbed since acquiring the remanence so that its direction is still meaningful. Hence, it is mostly red structural features that are suitable for analysis. Remains of furnaces and kilns are best suited. These are typically composed of clay, tile or brick, all of which usually contain suitable magnetic minerals. Furthermore, during their operation, these features reach temperatures in excess of 700 C, above the Curie temperatures of all the remanence-carrying minerals. For example, gure 5 shows the base of a Roman kiln constructed of red clay discovered at Heybridge in Essex. However, it is not always necessary for such high temperatures to be reached and the remains of domestic hearths and ovens can often be dated, even though they tend to possess weaker magnetizations. The example shown in gure 6 is a medieval hearth composed of ironstone from Burton Dasset in Warwickshire. Similarly, burnt or heated natural soil that has lain beneath a re or red structure can also be suitable in some instances. Dates have been obtained from the red clay soil beneath Tudor glass-making furnaces at Bagots Park in Staffordshire (gure 7). At this site, the remains of the furnaces themselves had been removed in the 1960s to allow the area to be ploughed. With all thermoremanent features it is
150
PHYSICS EDUCATION

Figure 6. Medieval hearth constructed of vertically stacked tiles at Burton Dasset in Warwickshire. Archaeomagnetism demonstrated that it was last used at the time of the documented abandonment of the settlement in the late fteenth century.

important to bear in mind that each time they are red their magnetization will be reset. Hence, the event dated by archaeomagnetic analysis will be the nal ring of the feature. Features possessing depositional remanence are less commonly encountered. However, where a waterlogged ditch has lled due to slow accumulation of sediment, such as the example shown in gure 8 from Yarnton in Oxfordshire, it is sometimes possible to date the time at which the sediment accumulated. Here, it was established that a prehistoric drainage ditch fell out of use and silted up during the Iron Age. Pictured in gure 9 is a Palaeolithic sequence of marine sands from Boxgrove in West Sussex. Here a stratigraphically related sequence of archaeomagnetic directions was obtained, showing the changes in direction of the Earths eld during the time over which the sediment layers accumulated (500 000200 000 BP). This sequence is too old to date by comparison with present UK calibration data but it has established that the site is less than 780 000 years old (when the last magnetic polarity reversal occurred). With all depositional remanent
March 2004

Archaeomagnetic dating

Figure 7. Fired clay soil originally beneath a Tudor glassmaking furnace at Bagots Park in Staffordshire. Archaeomagnetic dates on soils beneath 15 such furnaces have contributed to knowledge of the economics of glassmaking in 16th century England.

Figure 9. Marine sediment sequence laid down during the Palaeolithic period at Boxgrove in West Sussex. Sampling from different heights within the sequence has revealed a history of magnetic eld changes over thousands of years.

often tens of years after they rst settled out of suspension and can be much longer (Tarling 1983). Thus archaeomagnetism is most useful for dating older, prehistoric, sediments where such time lags are less signicant.

Sampling and measurement procedure


Since the direction of magnetization within an archaeological feature must be measured relative to true north and the horizontal plane, it is necessary to orient each sample before it is extracted or moved. For well consolidated features this is typically done by attaching a horizontally levelled marker to the sampling position with epoxy resin (gure 10). Levelling is achieved via a bulls-eye spirit level whilst the resin sets, then an arrow is marked onto the disk denoting the direction of true north. A gyro-theodolite can
PHYSICS EDUCATION

Figure 8. Ditch section at Yarnton in Oxfordshire. Archaeomagnetic analysis showed that the sediment lling the ditch accumulated between 200 and 100 BC, indicating that it had fallen out of use by this time.

magnetizations the event being dated is the time when the sediment particles became locked into position within the sediment column. This is
March 2004

151

P Linford

are usually sampled by enclosing a short pillar of sediment within a specially manufactured plastic cylinder which can be oriented as above, then removed and sealed. Ten to twenty samples must be extracted from different parts of each feature to be dated, to average out random perturbations in the recorded magnetization direction caused by material inhomogeneity and other factors. Laboratory measurement is usually carried out using a spinner magnetometer in which samples are spun within a pickup coil or ring uxgate (gure 11). In such magnetometers the sample is placed in a magnetically shielded measurement chamber to exclude the inuence of external magnetic elds. The sample sits on a platform atop a shaft, which is then turned at a xed speed to rotate the sample about its vertical axis. The rotating magnetic eld caused by the samples magnetization generates an electrical current in collecting coils wound around the measurement chamber, using the same principle as a dynamo. The magnitude of the current generated is proportional to the strength of the samples magnetization in the horizontal plane. By remeasuring the sample in three orthogonal orientations it is possible to determine the total direction and strength of its magnetization. For rapid measurement of very weakly magnetized samples, palaeomagnetic laboratories employ cryogenic SQUID magnetometers. Introduction of a magnetized sample into a superconducting ring causes a persistent current to ow that is proportional to the magnetic moment along the axis of the ring. Again, three measurements in orthogonal orientations are needed to completely determine the magnetization direction. More information about these and other types of magnetometer can be found in Collinson (1983).
Figure 10. Sampling archaeomagnetic features. Horizontally levelled markers are attached to the materials to be extracted in the top picture. In the bottom picture, the true north direction is being transcribed onto each marker with the aid of a gyro-theodolite.

Dating precision and limitations


With the present calibration data for the UK it is possible to date archaeological features to within about 50 years with 95% condence. However, the potential precision of the technique varies with the rate at which the magnetic pole position was moving at different periods. Where movement was rapid, features can be dated to within a smaller time window than in periods where movement was slow. This places a limit on the relative precision with which the dates of features from different periods can be determined, regardless
March 2004

be used for this (a device capable of nding the direction of true north using the precession of a built-in gyroscope), although a magnetic or sun compass bearing will often sufce. Once these procedures have been completed the sample can be removed. Less well consolidated sediments
152
PHYSICS EDUCATION

Archaeomagnetic dating

ring fluxgate

sample

reference signal

motor

Figure 11. Spinner magnetometer used to measure the magnetization within a sample by placing it on a rotating platform inside a measuring coil or ring fluxgate. The current generated by the rotating magnetic dipole within the sample will be proportional to the strength of its magnetization.

of the quality of the calibration evidence. In addition, it also appears that the magnetic pole has reoccupied the same positions at different times over the last 3000 years. This means that the calibration curve crosses itself and can lead to uncertainty as to which of two dates is correct for a particular remanence direction. At present, consideration of related archaeological evidence is often used to resolve such dilemmas, but it is hoped that current research will allow eld intensity measurements to be used to distinguish between the possibilities in future.

also been used to determine the provenance of obsidian (McDougall and Tarling 1983) and ceramics (Rasmussen 2001) and remanence and other magnetic properties have been used to determine the ring temperatures of pottery (Coey et al 1979). This is far from an exhaustive list of the possibilities that have already been investigated and new applications of archaeomagnetism will doubtless emerge in the future.

Related websites Conclusion


This article has briey overviewed magnetism as an archaeological dating tool; however, of necessity, much information has been omitted. More detailed discussions can be found in Aitken (1990) and, at a more technical level, in Tarling (1983). Magnetism is also used in other ways in archaeology. The use of portable magnetometers as a prospecting tool is now well established in British archaeology, and magnetic analysis of sediments is increasingly being applied to study environmental changes in antiquity (Thompson and Oldeld 1986). Magnetic properties have
March 2004

C M Batt: AARCHArchaeomagnetic Applications for the Rescue of Cultural Heritage www.bradford.ac.uk/acad/archsci/aarch/ G A Glatzmaier: The Geodynamo www.es.ucsc.edu/glatz/geodynamo.html S Lengyel: North American Archaeomagnetism www.u.arizona.edu/slengyel/ amag2 new-sw 001.htm R Nave: Magnetic Field of the Earth hyperphysics.phy-astr.gsu.edu/hbase/ magnetic/magearth.html
PHYSICS EDUCATION

153

P Linford

Scientic American: Ask the Experts: Geology: Why does the Earths magnetic eld ip...? www.sciam.com/askexpert question.cfm ?articleID=00010D87-BB04-1C71-9EB78 09EC588F2D7&pageNumber=1&catID=3
Received 14 October 2003, in nal form 19 December 2003 PII: S0031-9120(04)70307-X DOI: 10.1088/0031-9120/39/2/002

References
Aitken M J 1990 Science-based Dating in Archaeology (London: Longman) chapter 9 Clark A J, Tarling D H and Noel M 1988 Developments in archaeomagnetic dating in Britain J. Arch. Sci. 15 64567 Coey J M D, Bouchez R and Dang N V 1979 Ancient techniques J. Appl. Phys. 50 77727 Collinson D W 1983 Methods in Palaeomagnetism and Rock Magnetism (London: Chapman and Hall) Hoffman K A 1988 Ancient magnetic reversals: Clues to the geodynamo Sci. Am. 285 509

McDougall I and Tarling D H 1983 The magnetic sourcing of obsidian samples from Mediterranean and Near Eastern sources J. Arch. Sci. 10 44152 Nel L 1955 Some theoretical aspects of rock magnetism Adv. Phys. 4 191243 Rasmussen K L 2001 FOCUS: Provenance of ceramics revealed by magnetic susceptibility and thermoluminescence J. Arch. Sci. 28 4516 Tarling D H 1983 Palaeomagnetism: Principles and Applications in Geology, Geophysics and Archaeology (London: Chapman and Hall) Tauxe L 2002 Paleomagnetic Principles and Practice (Dordrecht: Kluwer Academic) Thellier E and Thellier O 1959 Sur lintensit du champ magntique terretre dans le pass historique et gologique Ann. Gophys. 15 285376 Thompson R and Oldeld F 1986 Environmental Magnetism (London: Allen & Unwin)
Paul Linford has worked for the English Heritage Ancient Monuments Laboratory, now part of the Centre for Archaeology, for 17 years and is joint head of the Geophysics Team. His specialisms are archaeomagnetic dating, electrical and magnetic prospecting and geophysical data processing.

154

PHYSICS EDUCATION

March 2004

S-ar putea să vă placă și