archaeomagnetic dating.ps, page 1-33 @ normalize ( 0406...

Archaeomagnetic DatingGuidelines on producing and interpreting archaeomagnetic dates

2006

0406 - Archaeomagnetic Dating 24/3/06 10:37 am Page cov3

2

Contents

Introduction 2

Part 1An introduction to archaeomagnetism 3The Earth's magnetic field 3Remanent magnetism 4Archaeomagnetic dating techniques 5What can be dated? 6Sampling procedure 7Laboratory measurement 8The mean remanent direction 9Dating precision and limitiations 10Future developments 13Other applications of archaeomagnetism 13

Part 2Practicalities: interactions between user and practitioner 14Planning 14Fieldwork 14

Suitability 14Safe working practice 14Sampling 15

Assessment of potential for analysis 15

Analysis and report preparation 16The archaeomagnetic dating report 16

Dissemination 16Summarising archaeomagnetic dates

in publications 18Citing archaeomagnetic dates 18

Data archiving 18Case studies 18

Dogmersfield House brick kiln (thermoremanent magnetisation) 18

Yarnton ditch sediment (post-depositional remanent magnetisation) 20

Glossary 23

Appendix 1Contact addresses for UK archaeomagnetism 26English Heritage 26Other sources of advice 27

Appendix 2Brief notes on the history of archaeomagnetism 28

Bibliography 28

Introduction

Scientific dating methods are wellestablished as important tools inarchaeological chronology and theirapplication as part of archaeologicalinvestigations is now routine. Quite anumber of techniques are available andperhaps the best known and most heavilyutilised are radiocarbon dating anddendrochronology. However, both requirethe preservation of sufficient quantities oforganic materials and, when this is notavailable, different dating methods arerequired. One such is archaeomagneticdating which has the attraction that it can directly date fired ceramic and stonematerials which frequently occur inarchaeological contexts. It is applicable to a more limited class of archaeologicalremains than radiocarbon dating althoughresearch is under way to redress this limitation (see below Futuredevelopments). Fortunately the periodswhere archaeomagnetic dating has thepotential to be most precise coincide with those where radiocarbon dating is problematic, making it a valuablecomplement to the latter technique.

A long history of investigation underliescurrent archaeomagnetic practice and

as a result of this research, the techniquehas developed to become a useful tool forarchaeological chronology so that it is nowpossible to date archaeological structuresin the United Kingdom from the last threemillennia. However, the precision of thedates that can be obtained varies fordifferent periods according to a number of factors (see below Dating precision and limitations).

As the result of a more stringent planningpolicy (Department of Environment 1990),greater emphasis has been placed on the protection and recording of thearchaeological resource when newdevelopment is planned. This has led to an increase in the number ofarchaeological sites being investigated and thus to the discovery of many morefeatures suitable for archaeomagneticanalysis. The purpose of these guidelines,therefore, is to provide information aboutthe technique of archaeomagnetic datingand the types of archaeological feature for which it is suited.

Since it is not possible to exploitarchaeomagnetic analysis to its fullpotential without an insight into the

principles upon which it is based, Part 1of this document provides an introductionto the theory behind it and how this hasbeen translated into a practical datingmethod. This part is primarily aimed atthose archaeological professionals who are unfamiliar with the technique.

Those wishing to use archaeomagneticdating should find Part 2 most valuable,especially the sections dealing withplanning and fieldwork. Archaeomagneticpractitioners, particularly newcomers tothe field, should also find Part 2 useful inestablishing a code of good practice whendealing with clients as this part discussesthe practical issues of how archaeomagneticdating may be used in an archaeologicalproject. It includes advice on the processesinvolved, from planning to the disseminationof results. Part 2 also suggests guidelinesfor reporting archaeomagnetic results toensure consistency and allow for their reusein the future as the technique develops.Two case studies illustrate how theprinciples and techniques are applied inpractice. Later sections provide a glossary of specialised meanings, useful contactsfor information and advice, and brief noteson the history of archaeomagnetism.

0406 - Archaeomagnetic Dating 24/3/06 10:37 am

Part 1

An introduction toarchaeomagnetism

Archaeomagnetism depends upon twoimportant physical phenomena:

1 The Earth spontaneously generates amagnetic field which changes in bothintensity (strength) and direction withtime (secular variation).

2 Specific events can cause naturallyoccurring magnetic minerals to becomepermanently magnetised, recording themagnetic field pertaining at the time of the event.

The next two sections expand upon eachof these phenomena in turn and describehow they are exploited to develop a usefuldating technique.

The Earth’s magnetic field The details of the mechanism by whichthe Earth’s magnetic field is generated are not completely understood. It appearsto be associated with a region, 3000kmbeneath the planet’s surface in the outercore, which is mostly composed of slowlychurning molten iron.This layer is trappedbetween the solid inner core at the centreof the planet and the mantle, another solidlayer extending from 3000km to about40km beneath the surface. It is nowgenerally accepted that free electroncirculation within the convecting outercore creates the magnetic field, whichbehaves as a self-sustaining dynamo (see, for instance, Merrill et al 1996).The fluid motions that drive this dynamoderive from the Earth’s rotation along with gravitational and thermodynamiceffects in and around the core.

This results in an approximately dipolarmagnetic field at the Earth’s surface, as if a large bar magnet was situated at itscentre with its long axis aligned almostparallel with the Earth’s rotational axis(Fig 1). Near the equator, the field lines(which indicate the direction in which afreely rotating magnetised needle wouldpoint) are directed horizontally, parallel to the surface of the Earth; however, aseither of the two magnetic poles isapproached 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 Earth’smagnetic poles do not exactly coincidewith its geographic (rotational) poles and

presently the axis of the dipolar field isinclined at 11.5° to the Earth’s rotationalaxis. Because of this, the direction indicatedby a compass needle at an arbitrary pointon the Earth’s surface will generally deviatefrom the direction of geographic, or true,north. The angle in the horizontal planebetween magnetic north and true north is known as the magnetic declination.

Over geologic timescales the position ofthe magnetic poles appears to precess aboutthe geographic poles and this is an effectof the same forces that generate the Earth’sfield. Superimposed upon this generallycirculatory movement is an apparentlyrandom element, believed to be due toeddy currents in the Earth’s fluid core and to the movement of charged particlesin the upper atmosphere. It is not only the position of the magnetic poles thatchanges but also the strength or intensityof the magnetic field. The field intensitydetermines the strength of the attractionof a compass needle to the magnetic poles and the strength of magnetisationacquired by magnetic minerals. Over thelast 150 years observations in London andParis indicate that the Earth’s field haschanged in direction by about 0.25° andin intensity by about 0.05% each year(Tarling 1983, 146). These long-termchanges in the terrestrial magnetic field,known as secular variations, form the basisfor the archaeomagnetic dating technique.Provided that a data base of secular

variation over archaeological timescalescan be built up, objects recording thestrength or direction of the Earth’s field(see below) can be dated by comparisonwith it.

Unfortunately, some of the causes of changeto the Earth’s magnetic field are localisedin their influence (the eddy currents andatmospheric movements). As a result,the magnetic intensity and pole positionscalculated from measurements made atone point on the Earth’s surface will notexactly match those calculated frommeasurements made at another position.For archaeomagnetic dating, this has theconsequence of requiring the compilationof separate calibration data bases of thesecular variation for different regions,each about 1000km in diameter.

One additional interesting aspect of thevariation in the intensity of the Earth’sfield has been discovered through studiesof the magnetisations recorded in igneousrocks. It appears that every few hundredthousand years the field intensity decreasesalmost to nothing then increases again withthe polarity reversed, that is, the northand south poles change places (Hoffman1988). Although of limited applicabilityfor archaeological dating, early hominidremains have been dated by counting thenumber of magnetic reversals recorded in the sediments deposited above them(Partridge et al 1999).

3

Angle of inclination

Magneticnorth pole

Geographicnorth pole

Currents generatingmain dipole field

Eddy current

Mantle

Fluid outer core

SN

Fig 1 The Earth’s 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 magnetic located at the Earth’scentre 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 Earth's surface.


Remanent magnetismSome naturally occurring minerals arecapable of retaining a permanent orremanent magnetisation. By far the mostprevalent of these are the iron oxides – magnetite, maghaemite and haematite –which occur in most soils, clays and astrace components in many types of rock(Thompson and Oldfield 1986). In crystalsof these minerals, quantum mechanicalexchange interactions between neighbouringatoms force all the unpaired electronmagnetic moments to align, resulting in a net spontaneous magnetisation (Néel1955; Tauxe 2002). Such alignment willoccur within a region of the crystal knownas a magnetic domain and the shape andsize of these domains will depend on thestructure and size of the crystal as well as the impurities within it. Each maycorrespond to one physical grain of themineral crystal (single domain), or a single grain may be divided into severalmagnetic domains to lower the overallenergy of the system (multidomain).Each magnetic domain will have one or more preferred directions ofmagnetisation, or easy axes, determined by its shape and underlying crystalstructure. Magneto-crystalline energy is minimised when the magnetisationwithin the domain lies in one of the two directions parallel to an easy axis,so the domain will tend always tomagnetise in one of these favourabledirections.

As each domain will typically bemagnetised in a different, randomlyorientated, direction, a macroscopicsample of the mineral containing a largenumber of domains will usually exhibit a negligible net magnetisation. However,if the mineral is heated, thermal agitationof the crystal structure leads to a diminutionof the spontaneous magnetisation in each domain until, at a certain criticaltemperature, known as the blockingtemperature, it disappears entirely (Fig 2).On cooling, each domain will remagnetisein the direction of the easy axis mostclosely parallel to any ambient magneticfield direction. Although most individualdomains’ magnetisations will not be exactlyaligned with the ambient field direction,they will tend to favour it on average.Thus,after heating, the mineral will exhibit a netthermoremanent magnetisation (TRM) inthe direction of the prevailing magneticfield at the time it cooled.

The blocking temperatures of differentmagnetic domains vary. The maximumblocking temperature is limited by theCurie temperature of the particular mineralinvolved (585°C for magnetite and 675°Cfor haematite) but considerations of grainsize, crystal structure and purity canreduce it below this limit. Indeed evenwithout heating, and in the absence of an external magnetic field, some domainswill spontaneously lose their directionsover time. Generally, the probability that

such a change will occur during a set timespan depends on the domain’s blockingtemperature (the lower it is the morelikely it is that a change will take place).Naturally occurring samples of rocks andclays will usually contain a heterogeneousmineral and domain-size composition andthus exhibit a spectrum of blockingtemperatures. To be capable of retaining a stable TRM, they must contain a highproportion of magnetic domains withblocking temperatures about 200°C.Domains with blocking temperaturesbelow 200°C are likely to realign overarchaeological timescales even withoutheating and their directions will trackchanges in the Earth’s magnetic field.Thisphenomenon, called viscous remanentmagnetisation (VRM), can lead to a partialoverprinting of the TRM acquired whenthe sample was originally heated. Whenmeasuring archaeomagnetic samples, caremust be taken to identify and remove theeffects of such viscous remanences.

As well as TRM, a second mechanism,called depositional remanent magnetisation(DRM), also occurs in archaeologicalcontexts. This involves water-bornesediment particles that possess a weakoverall magnetisation, often because they are composed of thermoremanentlymagnetised minerals (although chemicaleffects during crystal growth can alsoresult in remanent magnetisation). Ifsuspended in relatively still water, they willattempt to rotate so that their directions of magnetisation align with the prevailingmagnetic field (Fig 3). Gravitational forceswill tend to pull the particles to the bed ofthe body of water where they will settle toform a layer magnetised in the direction of this ambient field. As more sedimentaccumulates above this layer, frictionalforces caused by its weight eventually lock the magnetised particles in place sothat they are no longer free to rotate andrealign themselves. Further sedimentaccumulation results in a stratigraphicsequence of magnetic layers, thus recordingchanges in the Earth’s magnetic field overthe duration of their deposition. In certaininstances the sediment particles may notbecome locked into position until sometime after they are deposited and it is also possible that chemical or other effects can modify the initial depositionalmagnetisation. In these cases the resultingmagnetisation is referred to as a post-depositional remanent magnetisation(pDRM). Such magnetised sediments can occur on lake beds and have also been found in palaeochannels and

4

Key

Ambient magnetic field

Induced magnetisation

a)

b)

c)

d)

e)

Fig 2 Thermoremanent magnetisation. Initially magnetic domains within a sample are magnetised in random directions thatcancel out (a). As the sample is heated the domains demagnetise as the temperature exceeds their blocking temperatures (b and c). On cooling, the domains remagnetise in a direction close to the prevailing ambient magnetic field, resulting in a net magnetisation within the sample (d and e).


archaeological ditch sections where the rate of water flow was relatively low.

Archaeomagnetic dating techniquesIf the history of the changes in the Earth’s magnetic field is known, thenthere are two principal ways to date anarchaeological artefact, structure ordeposit that has acquired a remanentmagnetisation at some time in the past.

The first method is to exploit the fact thatthe direction of the Earth’s magnetic fieldhas changed over time.When the Earth’sfield is recorded by a magnetic material as described above it is generally easier to deduce the field direction frommeasurements made on the material than it is to infer the field intensity. Hence,archaeomagnetic research in the UKespecially has concentrated on developingthe archaeodirectional technique whichinvolves only the direction of the Earth’smagnetic field. The technique involvesestablishing the apparent magnetic northpole position indicated by the declination

and inclination of the field within thesample and determining when in the pastlocal magnetic north was in that position.Clearly, the artefact must have remainedin exactly the same position as it was whenit acquired its remanence, limiting the typesof object that can be dated to non-portablestructures. To this end, analysis of lakesediment data (depositional remanence), alarge number of well-dated archaeologicalstructures (thermoremanence), as well asdirect compass measurements from thelast 400 years, has led to the constructionof the United Kingdom archaeomagneticcalibration curve (Clark et al 1988).

Based upon these data, Fig 4 shows thevariation in the apparent position of themagnetic north pole as viewed from the UKover the past 3000 years. In principle, thedate at which an unmoved archaeologicalobject acquired its remanence can beinferred by measuring the declination andinclination of its magnetisation, determiningthe corresponding magnetic pole positionand comparing this to the dated timeline.

The main problem with thearchaeodirectional technique is that if a magnetised object is moved, thedirection of magnetisation within it is no longer meaningful. Hence a secondarchaeomagnetic dating method has alsobeen developed that infers the intensity of the Earth’s field at the time that anaretefact acquired its remanence from the strength of the magnetisation in theartefact. As the direction of the field is not involved, this has the attraction that it allows portable archaeological objectssuch as potsherds to be dated. Thestrength of magnetisation acquired by a fired sample depends on the strength

5

-170-160

-150

-140

-130

-120

-110

-100

-90-80

-70-60

-50

-40-30

-20-10 0oE 10 20

3040

50

6070

80

90100

110

120

130

140

150

160170180 -170

-160

-150

-140

-130

-120

-110

-100

-90-80

-70-60

-50

-40-30

-20-10 0oE 10 20

3040

50

6070

80

90100

110

120

130

140

150

160170180

1200 BC

1000 BC800 BC

700 BC

600 BC

500 BC400 BC

300 BC

200 BC100 BC

0

AD 100

AD 150

AD 200

AD 250

AD 300

AD 400

AD 500

AD 600AD 800

AD 900 AD 1000

AD 1050

AD 1100

AD 1200

AD 1250AD 1300

AD 1350AD 1400

AD 1500

AD 1550

AD 1700

AD 1800

AD 1850AD 1900

AD 1950

AD 1750

Fig 4 Movement of the apparent position of the Earth’s magnetic north pole with time, based upon archaeomagneticmeasurements made in the UK.The left-hand diagram shows movement from the Bronze Age until AD 500, the right-handdiagram shows movement from then until the present.

a)

b)

c)

d)

of the Earth’s magnetic field at the time it was fired but it can be influenced by many other factors such as mineralogicalcomposition and firing temperature.A technique to correct for these otherfactors and to estimate the ambient fieldstrength at the time of firing was developedby E Thellier (1938) and has subsequentlyundergone a number of refinements.It involves repeatedly heating the sample ina controlled (or zero) magnetic field to anumber of increasing temperature stagesand measuring the intensity of themagnetisation remaining after each stage.Once all the magnetisation is removed, theprocess is repeated but with the sampleexposed to a known reference magneticfield during each heating. From acomparison of the results it is possible toinfer the strength of the Earth’s field at thetime the sample was originally heated inantiquity. However, many measurementsare required for this technique, ascompared with directional dating, andmore sources of error are involved.Thesecan arise from sample inhomogeneity,sample anisotropy, differences between theoriginal and laboratory firing atmosphereand differences between the original andlaboratory heating and cooling rates(Tarling 1983, 149).

Whilst archaeointensity studies have metwith success in a number of Europeancountries (see, for instance, Lanos et al1999; Kovacheva et al 2000), little workhas been undertaken in the UK owing tothe problems involved. Knowledge of theEarth’s past magnetic field intensity in thevicinity of the British Isles is based upononly six studies, none of which was madewith modern equipment or methodologies.However, as outlined below in Future

Fig 3 Depositional remanent magnetisation. Sedimentparticles, each with a weak magnetisation, settle out of stillwater. As they fall through the water column they rotate to align their internal magnetisation directions with theEarth’s magnetic field (a, b, and c). Once settled on the bedof the body of water, the weight of sediment accumulatingon top of the particles locks them in place, leaving a layermagnetised in the direction of the Earth’s field (d).


developments, a new demagnetisationtechnique shows great promise and islikely to stimulate renewed interest in UK archaeointensity studies over the next few years.

What can be dated?Given the paucity of archaeointensitycalibration data for the UK, thearchaeointensity technique is at presentunlikely to be encountered except in aresearch context for English archaeologicalfeatures. Hence, the following sectionsconcentrate on the archaeodirectionaltechnique which is sufficiently welldeveloped in the UK for a dating service to be available.

Directional archaeomagnetic datingimposes three constraints on the types ofarchaeological features that can be dated.They must:

1 contain magnetic minerals capable of carrying a stable remanentmagnetisation;

2 have experienced a remanence-inducing event at some time in theirhistory, for example, heating above ablocking temperature or non-turbulentsediment deposition;

3 have remained undisturbed sinceacquiring the remanence so that themagnetisation directions they recordare still meaningful.

Hence, it is mostly fired structural featuresthat are suitable for analysis. Remains offurnaces and kilns are best suited.These are typically composed of clay, tile, brickor stone, all of which usually containsuitable magnetic minerals. Furthermore,during their operation, these featuresreach temperatures in excess of 700°Cabove the Curie temperatures of all theremanence carrying minerals. For example,Fig 5 shows the base of a Roman kilnconstructed of fired clay discovered atHeybridge, Essex (Noël 1996). However,it is not always necessary for such hightemperatures to be reached and the remainsof domestic hearths and ovens can oftenbe dated, even though they tend to possessweaker magnetisations.The example shownin Fig 6 is a medieval hearth composed of ironstone from Burton Dassett,Warwickshire (Linford 1990). Similarly,burnt or heated natural soil that has lainbeneath a fire or fired structure can alsobe suitable in some instances. Dates havebeen obtained from the fired clay soilbeneath medieval and Tudor glass-makingfurnaces at Bagot’s Park, Staffordshire(Fig 7). At this site, the remains of thefurnaces were removed in the 1960s toallow the area to be ploughed (Linfordand Welch 2004). Although these examplesare predominantly fired horizontal surfaces,burnt walls (eg kiln walls) can also be dated when they survive and have notcollapsed or moved since the firing event.Furthermore, whilst the majority of featuresthat are dated archaeomagnetically arecomposed of clay or ceramic materials,it should be emphasised that burnt stonestructures are also often suitable. Even stonetypes not usually associated with ironminerals, such as limestone, can oftencontain trace quantities of magnetic mineralscapable of retaining a magnetic remanence.

With all thermoremanent features it is important to bear in mind that eachtime they are fired their magnetisation will be reset. Hence, the event dated byarchaeomagnetic analysis will be the finalfiring of the feature. However, a caveat tothis restriction can occur in the case wherethe final heating of a feature was to a lowertemperature than reached in a previousfiring. In such instances, it may be possibleto date both firings.

Features possessing depositional remanenceare less commonly encountered. However,where a waterlogged ditch has filled dueto slow accumulation of sediment, such asthe example shown in Fig 8 from Yarnton,Oxfordshire, it is sometimes possible to

6

Fig 5 (top) Base of a Roman pottery kiln (~1.5m in diameter), constructed of fired clay, discovered at Heybridge,Essex and dated to the 2nd century AD.

Fig 6 (bottom, left) Medieval hearth constructed of vertically stacked tiles at Burton Dassett,Warwickshire.Archaeomagnetism demonstrated that it was last used at the time of the documented abandonment of the settlement in the late 15th century.

Fig 7 (bottom, right) Fired clay soil originally beneath a Tudor glassmaking furnace at Bagot’s Park, Staffordshire.Archaeomagnetic dates on soils beneath 15 such furnaces have contributed to the knowledge of the economics ofglassmaking in 16th-century England.


date the time at which sedimentationoccurred. In this case, it was establishedthat a prehistoric drainage ditch fell out of use and silted up during the Iron Age(Linford et al 2005). Pictured in Fig 9 is a Middle Pleistocene sequence of marinesands from Boxgrove, West Sussex (Davidand Linford 1999). Here a stratigraphicallyrelated sequence of archaeomagneticdirections was obtained showing thechanges in direction of the Earth’s fieldduring the time over which the sedimentlayers accumulated (~500,000–200,000 BP).This sequence is too old to date bycomparison with present UK calibrationdata but it confirms that the site is less than780,000 years old (when the last magneticpolarity reversal occurred). Using the sametechnique, analysis of a similar sedimentsequence from Gran Dolina, Spain hasdemonstrated that hominid remains aremuch earlier than first supposed, dating to before the last geomagnetic reversal(Gutin 1995).

For depositional remanences to occur, thebody of water from which sedimentationis taking place needs to have a slow rate of flow. Hence, lake and pond sedimentsare often well suited as are palaeochannelsthat become cut off and then silt up.Low-energy flood deposits have also beendated. Subsequent bioturbation (eg bytree roots) can disturb the sediment andrender the remanence undatable but smallparticles of organic matter deposited atthe same time as the sediment do notnecessarily affect the locking-in process.It should be noted that with all DRMs the event being dated is the time when the sediment particles became locked

into position within the sediment column.A sufficient accumulation of sediment isrequired above the layer in question tocause adequate compaction. Whilstsediments composed of fine-grained claysmay be locked in almost simultaneouslyduring ongoing sedimentation, coarsergrained silts such as loess sediments may not be locked in until several metresof sediment have accumulated above them(Evans and Heller 2003, 86). Clearly, thetime required for a sufficient weight ofsediment build-up will also depend on the rate of sedimentation. This uncertaintime-lag between sediment deposition andlock-in means that it is often difficult toassociate depositional remanences with an archaeological event and Batt (1999)cautions that this presently poses asignificant obstacle to dating sedimentsusing archaeomagnetism. For this reason,depositional archaeomagnetism is typicallyuseful for dating older, prehistoric sedimentswhere such time-lags may be less significant.

Sampling procedureSince the direction of magnetisation withinan archaeological feature must be measuredrelative to true north and the horizontalplane, it is necessary to orient each samplebefore it is extracted or moved. For wellconsolidated features this is usually doneby creating a flat surface on the material tobe sampled upon which a direction arrowcan be marked.To date, the most commonmethod for achieving this employed byUK practitioners has been to attach, withepoxy resin, a horizontally levelled plasticmarker disc at the sampling position (Fig 10). This is levelled using a bull’s eyespirit level while the resin sets to ensure

7

Fig 8 (above) Ditch section at Yarnton, Oxfordshire. Archaeomagnetic analysis showed that the sediment filling the ditchaccumulated between 200 and 100 BC, indicating that it had fallen out of use by this time.

Fig 9 (right, top) Marine sediment sequence laid down during the Middle Pleistocene period at Boxgrove,West Sussex.Sampling from different heights within the sequence has revealed a history of magnetic field changes over thousands of years.

Fig 10 Sampling consolidated features. Horizontally levelledmarkers are attached to the materials to be extracted in themiddle picture. In the bottom picture, the true north directionis being transcribed onto each marker with the aid of agyro-theodolite.


each other, it is possible to determine the total direction and strength of itsmagnetisation.

For rapid measurement of very weaklymagnetised specimens, cryogenic SQUIDmagnetometers can be used. Introductionof a magnetised specimen into asuperconducting ring causes a persistentcurrent to flow that is proportional to themagnetisation parallel to the axis of thering. Again, three different measurementorientations are usually needed tocompletely determine the magnetisationdirection. Typical archaeomagneticspecimens tend to be relatively stronglymagnetised and the spinner magnetometeris usually the most appropriatemeasurement instrument. However,SQUID magnetometers are employed in research studies, particularly whenmeasuring weakly magnetised sedimentsor where only very small samples could be collected (eg from fired clay artefacts).More information about the various typesof magnetometer can be found inCollinson (1983).

The accumulation of VRM in a magneticmaterial left undisturbed in a magneticfield for a long period of time has beenreferred to above in the section onRemanent magnetism. The new viscousremanence partially overprints the originalmagnetisation, altering the measurementsof magnetisation direction and intensitymade on untreated specimens in thelaboratory.Thus, it is desirable to removethe viscous remanence whilst causing aslittle change as possible to the primarymagnetisation. This is done by partiallydemagnetising the specimen, exploitingthe fact that the viscous magnetisation will be carried by the magnetic domainsof lower stability. One of two methods is typically employed: either heating thespecimen (thermal demagnetisation) orexposing it to an alternating magneticfield (AF demagnetisation).

Thermal demagnetisation involvesheating the specimen in a zero fieldenvironment to a temperature above the blocking temperature of the viscousdomains but below that of the stabledomains carrying the remanence ofinterest. On cooling, the magnetisationdirections of all domains with blockingtemperatures below the thresholdtemperature will have been randomised.By successively reheating the specimen tohigher temperatures and measuring thechange in the direction of the sample’s

its top surface is horizontal. However, anappropriate surface may also be preparedeither by simply skimming the surface ofthe material to be sampled itself or byattaching plaster to it then flattening theplaster’s top surface as it dries. In thesecases, the flat surface created is not alwayshorizontal. Instead, an inclinometer isused to determine the degree and directionof the surface’s slope.

Whichever method is used to create a flat sampling surface, an arrow must bemarked onto it denoting an accuratelyestablished reference direction, typicallytrue north. Usually, a sun compass orgyro-theodolite is used to establish thereference direction. However, a magneticcompass bearing can be employed insituations where there is little localiseddisturbance to the magnetic field (whichcan be caused by nearby ferrous structuresor by the feature to be dated being stronglymagnetised). Once these procedures have been completed the sample can be

removed. Less well consolidated sedimentsare often sampled by enclosing a short pillarof sediment within a specially manufacturedplastic cylinder (Fig 11), which can beoriented as above, then removed and sealed.Regardless of the method of sampling 10 to 20 samples must be extracted fromdifferent parts of each feature to be dated to average out random perturbations in therecorded magnetisation direction caused bymaterial inhomogeneity and other factors.

Laboratory measurementBefore measurement, friable samples areoften treated with a consolidant such as PVA(polyvinyl acetate) in acetone or sodiumsilicate to ensure they do not fragmentduring measurement. Subsequently, wherelarge samples have been collected from afeature, or where each sample is an entirebrick or tile, they are often sub-sampled in the laboratory to produce a set ofspecimens. The magnetisation of eachspecimen is measured individually thenthe measurements of all specimens takenfrom a particular sample can be averagedto produce a mean magnetisation for thesample. The advantage of this approach is that poor samples, where the magneticmaterial does not record a consistentmagnetisation direction, can be readilydetected and rejected from further analysis.The disadvantage is that much largerquantities of material usually have to beremoved from the archaeological featureto be dated. Where smaller samples havebeen extracted (as is typical with the discmethod), a single measurement is oftenmade on each entire sample instead ofaveraging measurements on a number of specimens taken from it.

Laboratory measurement of magnetisationis usually carried out using a spinnermagnetometer in which specimens arespun within a pickup coil or ring fluxgate(Fig 12). In such magnetometers thespecimen is placed in a magneticallyshielded measurement chamber to excludethe influence of external magnetic fields.The specimen sits on a platform atop ashaft, which is then turned at a fixedspeed to rotate it about its vertical axis.The rotating magnetic field caused by the specimen’s magnetisation generates an electrical current in collecting coilswound around the measurement chamber,using the same principle as a dynamo.The magnitude of the current generated is proportional to the strength of thespecimen’s magnetisation in the horizontalplane. By re-measuring the specimen in three orientations at right angles to

8

Fig 11 Sampling unconsolidated features. (a) 10cc perspexcylinders that can be pushed, open end first, into sediments.They have an arrow marked on the base, so the cylindercan be rotated to align with the fiducial direction. A close-fitting lid closes the open end of the cylinder after it isexcavated. (b) Larger cylinders, with both ends open, thatcan be fitted over excavated monoliths of sediment.Thesediment is then sealed in with plaster of Paris.

a)

b)


Motor

ReferenceSignal

Ring Fluxgate

Sample

magnetisation at each step, the temperaturenecessary to completely remove the VRMcomponent can be established.

AF demagnetisation uses a weakalternating magnetic field instead ofheating the specimen. Generally, magneticdomains with low blocking temperatureswill also have low coercivities, so theirmagnetisation directions will move totrack the alternations of the applied field. The stable domains carrying thethermoremanence of interest will havehigh coercivities and the forces induced by the applied alternating field will be tooweak to alter their magnetisation directions.The peak strength of the alternating fieldis then slowly reduced to zero, leaving themagnetisation directions of the domainswith low coercivities randomised. As withthermal demagnetisation a succession of increasing peak alternating fieldstrengths can be used to determine theoptimum value for removal of the viscouscomponent.

Thermal demagnetisation has the advantagethat it is similar to the process that causedthe initial TRM to be acquired (heating)but repeated heating and cooling of thespecimen can cause chemical changes tothe magnetic minerals being measured.AF demagnetisation does not causechemical changes but there is someevidence to suggest that magnetic domainsdo not react to AF demagnetisation inexactly the same way as they do to thethermal changes which magnetised them in the first place. More informationabout techniques of demagnetisation can be found in Collinson (1983).

Typically, a specimen’s magnetisation will bere-measured after partial demagnetisationat a sequence of increasing temperatures orpeak field strengths. The succession of changes in direction and strength ofmagnetisation are then statistically analysedusing principal components analysis(Kirshvinck 1980) to determine theoptimal direction of magnetisation for the specimen. If the changes at each stageare too great, the magnetisation in thespecimen may be ruled unstable, in whichcase the specimen would be excluded fromthe next stage of analysis, the calculationof the mean remanence direction for the feature.

The mean remanent directionIf several specimens from the same featureare measured, it will be found that theirremanent magnetisations differ slightly inboth direction and intensity. In part thiswill be due to random errors introducedby the sampling and measurement process.Tarling (1983) estimates that whensampling and using a sun compass,errors of orientation should be within 2°.Empirical evidence suggesting that thisestimate is likely to be correct has beenobtained by Hathaway and Krause (1990)who compared azimuthal directionsmarked on samples from a number ofexperimental hearths orientated usingboth magnetic and sun compasses.The standard deviations for differencesbetween the two measurements averagedover all samples from a particular hearthwas typically of the order of 1° (thus within2° for 95% of the samples). Provided the intensity of remanence of specimensexceeds the noise level and sensitivity

9

Fig 12 Spinner magnetometer used to measure the magnetisation 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 beproportional to the strength of its magnetisation.

of the instrument, all types ofmagnetometer should be capable ofmeasuring magnetisations with arepeatability of some 0.5–1.0° for directionand 1–2% for intensity.Taking account ofthe need to make several measurementsafter different partial demagnetisation oneach specimen (and, if appropriate, tocalculate a sample average from severalspecimen magnetisation determinations),it should still be possible to determine thedirections of magnetisation of individualsamples to within about 2–3° and certainly no more than 5°. Intensities of magnetisation should be measurable to within about 5%.

However, the variation in remanenceobserved in a typical set of samples isoften greater than this (Tarling et al 1986)and the following other factors have beensuggested as additional causes of variationin the magnetisation within a feature:

1 The feature has been slightly disturbedsince it acquired its remanence causingdifferent parts of the feature to shiftslightly in different directions (this affectsthe magnetisation directions only).

2 Varying material composition within the feature. In the case ofthermoremanence, different magneticminerals have different blockingtemperatures and it is possible thatparts of the feature containing largequantities of minerals with highblocking temperatures were not heated sufficiently to fully realign the magnetisation directions (variableheating across the feature causessimilar effects).

3 Due to their composition, somemagnetic materials exhibit anisotropywhich means that they are easier tomagnetise in some directions than inothers. The magnetisation directionrecorded in such materials will tend to be distorted from its true valuetowards one of these more favourabledirections. This distortion can alsoaffect the apparent intensity of the field recorded.

4 The feature was in close proximity toan object with its own strong magneticfield when it acquired its remanence.This can occur in iron furnaces whichhave cooled with slag inside them.

5 Related to the above, and specific to TRMs, a feature composed ofstrongly magnetised material canexhibit distortions to the magnetic fieldrecorded within it due to its own shape.For instance, it has been noted that


may be possible to estimate and correctfor greater degrees of slumping if it can be assumed that the feature was levelwhen fired (Clark et al 1988).

Dating precision and limitationsThe precision with which a feature can bedated using directional archaeomagneticanalysis clearly depends inversely on the

the inclinations of samples taken fromthe walls of well-magnetised kilns are several degrees steeper than forsamples taken from the floors of thesame kilns. It has been proposed thatthis is due to the phenomenon ofmagnetic refraction (Aitken andHawley 1971; Schurr et al 1984) orthat it is due to the magnetisation ofthose parts of the feature that cool first,distorting the magnetic field throughthe feature (Tarling et al 1986), but thephenomenon is not well understood.

For the above reasons it is necessary totake a number of samples from all aroundthe feature to be dated and calculate amean remanent magnetisation directionfrom the magnetisations of the individualsamples. As has been described, thesamples originally extracted from thefeature may have been subdivided intoseveral specimens for analysis, in whichcase all the specimens from a particularsample will first be averaged to calculate a mean magnetisation direction for eachsample. Samples may be rejected from the calculation of the feature’s meanremanence direction at this stage if theirindividual specimen directions differwidely or if the changes in magnetisationduring partial demagnetisation indicatethat the remanence recorded is not stable.

Calculation of the mean direction ofremanence is performed by vector additionof the individual sample magnetisationdirections as shown in Fig 13. In thiscalculation the strength of the magnetisationof each sample is ignored as there is noreason to suppose that samples exhibitinghigh remanence intensity are more reliablethan those with weaker intensities. Hence,each sample’s magnetisation is representedby a vector of unit length. As the meanremanence direction is calculated from adistribution of sample magnetisations withdifferent directions, there is a degree ofuncertainty attached to the calculation. Astatistical method specific to the problemof analysing distributions of unit vectorsin three dimensions was developed by R A Fisher (1953) in which the uncertaintyis represented by the α95 (alpha-95)parameter. This is the semi-angle of acone of confidence centred on the meanvector direction within which there is a95% chance that the true mean vector ofthe distribution lies (see Fig 13c). It can be thought of as analogous to the standarderror in more familiar Gaussian statistics.The smaller α95 is, the more precisely themean direction is known and, in general,

the more precisely the feature can be dated.Once the mean vector and α95 statistic havebeen established, they can be comparedto the UK archaeomagnetic calibrationcurve to establish the date (or dates) whenthe Earth’s field had this direction. As the Earth’s field direction also varies withlocation, the comparison is usually doneby establishing the north pole positionindicated by the mean remanentmagnetisation direction, taking intoaccount the position of the site on theEarth’s surface. Such a pole position iscalled a virtual geomagnetic pole (VGP).The magnetic field declination andinclination that such a pole position wouldcause at Meriden (a central referencelocation for the UK) is then calculated(Tarling 1983, 116; Shuey et al 1970;Noël and Batt 1990). The correcteddirection can then be compared with theUK archaeomagnetic calibration curvewhich has been calculated at Meriden(Clark et al 1988; Tarling and Dobson1995; Batt 1997); a typical calibrationcomparison is depicted in Fig 13d.

It has been noted that the remains of amagnetised archaeological feature mightbe disturbed after the remanence hasbeen acquired and it is of interest todetermine the degree of displacement that can be sustained before it becomesundatable. The value of the α95 statisticvaries approximately inversely with thesquare root of the number of samplesused to calculate the mean magnetisationdirection, N (Tarling 1983, 121).Typically,N will be between about 10 and 18 and,as a rule of thumb, an α95 value of 2.5°or less is necessary for an adequatearchaeomagnetic date for most periodsusing the available UK calibration data.This suggests that the total directionalerror for an individual sample should not exceed about 10°. Hence, allowing for sampling and measurement errors ofthe order of 2–3°, deflections due to post-remanence acquisition disturbance mustnot exceed 7–8°. This assumes that thefeature breaks up and different partsmove randomly in different directions. Itmight be thought that taking an increasednumber of samples (ie increasing N)would allow greater degrees of disturbanceto be compensated for. However, a law ofdiminishing returns is rapidly encountered,as the analysis above assumes that eachsample position has moved randomly,and completely independently, of all othersample locations, and this condition rarelyholds in practice. Nevertheless, where theentire feature has moved as a whole, it

10

a)

b)

d)

900

1000

1100

1200

13001400

1500

1600

1700

1800

1900

circle ofconfidence

0 1020

30

-10

-20

80

70

60

50

-30

95

c)

α

Fig 13 The mean direction of remanent magnetisation.(a) The magnetisation directions of each sample will beslightly different. (b) A mean direction is calculated by vectoraddition of the individual sample directions (attaching eachsample direction vector to the end of the last). (c) Thecalculated mean direction is only an estimate of the truemean direction.The alpha-95 statistic, α95, describes thesemi-angle of a cone around the calculated mean directionwithin which there is a 95% probability that the true meandirection lies. (d) Comparing the circle of confidence of amean direction with an archaeomagnetic calibration curve(that of Batt 1997) using an equal angle stereogram plot.


magnitude of the α95 confidence statisticcalculated for the mean remanencedirection of the feature. Generally, thesmaller this angle, the more precisely the field direction can be established and thus the shorter the segment of thecalibration curve intersected.

The accuracy of the calibration data forthe period in question is equally important.The calibration curve has largely beenconstructed from archaeomagnetic analysisof features of known age and, in each case,the archaeomagnetic remanence directionhas been determined to a certain precisiongoverned by its own α95 statistic.Furthermore, there is also likely to be adegree of uncertainty in the independentdating evidence for the feature, resultingin the date the remanence was acquiredbeing known only within certain limits.Hence, the quantity and quality of theavailable archaeomagnetic calibration data will impose a limit to the precision to which a feature can be dated. With thepresent calibration data for the UK, thepractical limit on the maximum resolutionof dates is around 50 years at the 95%confidence level.

However, a third factor, the rate at whichthe VGP position changes, also governsthe precision of archaeomagnetic dates.The rate of change of the magnetic fielddirection has varied considerably over thelast 3000 years. In periods where movementwas rapid, features can be dated to withina smaller time window than in periodswhere movement was slow. This places afundamental limit on the relative precisionwith which the dates of features fromdifferent periods can be determined,regardless of the quality of the calibrationevidence or precision to which the meanremanence direction is known.

In addition to changes in the rate ofmovement, it also appears that the VGPhas reoccupied the same positions atseveral different times over the periodcovered by the UK calibration data.Hence, the calibration curve crosses itselfleading to uncertainty as to which of twodates is correct for a particular remanencedirection. Furthermore, given the limitson precision of archaeomagneticmeasurements, a similar situation occurswhen two segments of the calibrationcurve lie close to each other, even if they do not cross.

Since the late 1980s three archaeodirectionalcalibration curves have been published

for the UK and these are summarised in Table 1.1. It will be clear from the foregoingthat calibration curve construction is notstraightforward since it requires methodsto take account of uncertainties inmultivariate calibration data pointsdistributed unevenly over time (with adistinct paucity for some periods). Asstatistical and computational methodscapable of tackling the complexities of this process have been developed, therehas been an evolution from subjectiveevaluation of the available calibration data to a more objective assessment of the uncertainties involved and this isreflected in the curves cited in the table.It may be noted that the oldest feature thatmay presently be dated with reference to any of these curves would be around1000 BC. However, less precise estimatescan be obtained for earlier features by directcomparison with lake sediment data which

11

1000BC 700BC

600BC500BC

100BC0

100

200

400500

300BC

600

900

1000

1100

12001300

1400

1500

1700

1800

1900 1600

800

-30 -20 -10 0 10 20 30 40 50

50

60

70

80

Incl

inat

iono

-30 -20 -10 0 10 20 30 40 50

50

60

70

80

Declinationo

Declinationo

Incl

inat

iono

Fig 14 The directions of magnetic declination and inclination that would have been observed at Meriden in the centre of the UK over the past three millennia, based upon the data of Clark,Tarling and Noël (1988). Upper figure: 1000 BC to AD 550, lower figure: AD 550 to AD 1950. On both figures, dates with no suffix are AD, negative declinations are west of true north.

covers approximately the last 10,000 years(Thompson and Turner 1979;Turner andThompson 1981). As noted in the nextsection, Future developments, recentadvances are likely to result in a muchimproved UK calibration curve within the next few years.

The UK calibration curve of Clark,Tarling and Noël (1988), is depicted inFig 4 as a pair of VGP plots, one for theperiod up to about AD 500, and the otherfor the period from then until the presentday. The resultant variation in inclinationand declination of the Earth’s field thatwould have been observed in the centre of the UK is depicted in Fig 14. As abroad guide, the movement of the VGP as inferred from present calibrationinformation and the consequences forarchaeomagnetic dating have beengeneralised in Table 1.2.


Table 1.1 Archaeodirectional calibration curves for the UK published since the late 1980s

calibration curve date range covered description

Clark,Tarling and Noël (1988) 1000 BC to present Based upon ~92 calibration points from independently dated archaeological features as well as historical observations and lake sediment data. No assessment of the uncertainty inherent in the calibration curve was possible.

Tarling and Dobson (1995) 100 BC to present Based upon ~172 calibration points but did not use lake sediment data for prehistoric period.No objective assessment of uncertainty but stated to be no more that 5° for all periods.

Batt (1997) 1000 BC to present Used a similar database of calibration data to the above but lake sediment data were not included. Applied an objective moving window averaging method that provides an assessment of the uncertainty in the calibration curve at each point in time.

12

Table 1.2 Assessment of the potential of archaeomagnetic dating for the various broad periods of UK archaeology

period VGP movement potential for archaeomagnetic dating

Post medieval Rapid movement. First a rapid steepening of inclination Precision ~50 years at 95% confidence or better,AD 1485–present with constant declination between AD1450 and 1600. possibly ~20 years for dates after AD 1700.

Then a rapid westerly change in declination from AD 1600 to 1800. Finally an easterly change in Good potential owing to rapid VGP movement.declination to present. Inclination shallows between Excellent calibration data from AD 1570 when direct AD 1700 and present. observations began. Possible confusion with Iron Age

or early medieval dates around 17th century.

Medieval Rapid movement throughout period. Large westerly Precision ~50 years at 95% confidence.AD 1066–1485 swing in declination and drop in inclination. Apparent

‘loop’ between about AD 1280 and 1425. Good potential owing to rapid VGP movement Ambiguity with dates near AD 1280 and 1425 owing to a tight loop or possible crossover, also possibility of confusion with Roman dates during early 14th century.

Early medieval Slow movement from AD 400 to 850.Then more Precision ~100–200 years at 95% confidence, better AD 410–1066 rapid increase in declination. towards end of period.

Poor potential due to very slow VGP movement and present paucity of good calibration evidence. Possible confusion with Iron Age and 17th century AD for dates between AD 600 and 800.

Roman Relatively rapid drop in inclination between AD 100 Precision ~50 years at 95% confidence.AD 43–410 and 250. Inclination then increases again. Declination

fairly constant throughout. Potentially good precision between AD 100 and 300,but ~75–100 before and after this period. Double-backat AD 250 means independent evidence is often needed to determine if date is early or late Roman.Possible confusion with 14th century near AD 250.

Iron Age Rapid, linear, westerly change in declination with Precision ~70–100 years at 95% confidence.700 BC–AD 43 inclination relatively constant for most of the period.

Reversal of direction of change in declination around Potentially good but at present a paucity of calibration 50 BC causes ‘hairpin’ in 1st century BC. evidence limits precision. Hairpin in 1st century BC

can complicate dating at this period.

Bronze Age Very slow, almost stationary at a position with very Precision ~200 years at 95% confidence.2500 BC–700 BC easterly declination, and steep inclination.

Poor due to slow movement and paucity of calibrationinformation. Extreme position of VGP far from true north does allow features dating from this period to be easily distinguished.


Future developmentsResearch is continuing to improve thequality of the calibration data base forthose periods where the UK directionalarchaeomagnetic dating curve is presentlynot well defined. It is evident from Tarlingand Dobson (1995) that there is a lack ofgood quality calibration information forthe period before 100 BC as well as forthe period between AD 400 and 800.This results in less precise and less reliablearchaeomagnetic dates during these timeperiods. Whilst some headway might bemade by investigating ways of incorporatingexisting European data into the UKDatabase, more examples of independentlydated features from these periods areneeded to fully resolve this problem.

In tandem with gathering more calibrationdata, statistical research is being conductedto investigate better ways of compilingcalibration curves from the Database ofknown-age archaeomagnetic determinationsand then comparing magnetisation vectorsfrom features of unknown age with suchcurves. To date, UK calibration curveshave been constructed by visual inspection(Clark et al 1988), which is prone tosubjectivity, or by moving window averagingmethods (Batt 1997) that tend to dampenor flatten out real trends within the data(Lengyel and Eighmy 2002). Newadaptations of Bayesian statistics tospherical distributions promise to improvethe reliability and precision with whicharchaeomagnetic features can be dated(Lanos 2004). Furthermore, new techniquesfor modelling changes in the geomagneticfield over large regions promise to overcomethe need to correct all measurements to a central reference location and allowthe integration of reference data overmuch larger regions (Korte and Holme 2003).

However, perhaps the greatest weakness of archaeomagnetic dating in the UK isthe lack of calibration information forvariation in the intensity of the Earth’smagnetic field over past millennia. Datingusing the intensity as well as the directionof the Earth’s field could resolve many of the ambiguities caused by crossovers in the calibration curve discussed in theprevious section. One of the problems thathas hampered the development of a UKarchaeointensity calibration curve hasbeen the necessity for the use of thermaldemagnetisation to determine the strengthof magnetisation within each sample. Thisprocess can lead to chemical alteration of the magnetic minerals carrying the

remanence, thus changing the resultsobtained. However, a microwavedemagnetisation technique has beendeveloped that results in very little bulkheating of the samples being measured(Shaw et al 1996; Shaw et al 1999). Initialtests on archaeological material from theUK show promise (Casas et al 2005) and it is hoped that over the next fewyears research can be directed towardsconstructing a UK archaeointensitycalibration database using the newmicrowave technique.

To coordinate archaeomagnetic researchacross Europe and to address the presentshortage of trained archaeomagneticpractitioners, a European researchtraining network, ArchaeomagneticApplications for the Rescue of CulturalHeritage (AARCH), has been established.The network is being coordinated by Dr Cathy Batt at the University ofBradford. Information about thenetwork and its participants is available at the AARCH website:www.brad.ac.uk/acad/archsci/aarch/ or by contacting [email protected]

As part of this initiative, researchers in the network would be extremely interestedto be able to sample archaeomagneticfeatures for which independent datingevidence is also available. In the event of a suitable feature being discovered,contact Dr Cathy Batt or Paul Linford at English Heritage (see Appendix 1 forcontact details).

Other applications of archaeomagnetismAlthough this document focuses on theuse of archaeomagnetic analysis as adating technique, it can also provide othertypes of information to the archaeologist.Directional archaeomagnetic measurementcan determine which way up an objectwas when fired, test whether fired materialis in situ or if it has collapsed or beenredeposited and may even be able to help determine if sherds of a tile (or otherfired ceramic object) were once fittedtogether. Remanence and other magneticproperties have been used to estimate thefiring temperatures of pottery (Coey et al1979) and burnt sediments (Linford andPlatzman 2004) as well as the duration of firing experienced by hearths (Mengand Noël 1989). The use of portablemagnetometers as a prospecting tool isnow well established in British archaeology(Clark 1990; English Heritage 1995;Gaffney and Gater 2003) and magneticanalysis of sediments is increasingly being

13

applied to study environmental changes in antiquity (Thompson and Oldfield1986; Evans and Heller 2003). Magneticproperties have also been used to determinethe provenance of obsidian (McDougalland Tarling 1983), limestone (Williams-Thorpe and Thorpe 1993) and ceramics(Rasmussen 2001). This is far from anexhaustive list of the possibilities that have already been investigated and newapplications of archaeomagnetism willdoubtless emerge in the future.


Part 2

Practicalities: interactionsbetween user and practitioner

Archaeomagnetic analysis should be partof an integrated project framework. Theprocedures and principles of such projectmanagement should follow those set outin Management of Archaeological Projects(MAP2) (English Heritage 1991 andfamiliarity with Management of ResearchProjects in the Historic Environment(MoRPHE) (English Heritage forthcoming)is also advisable. The model for projectmanagement described in MAP2 iscomposed of six stages.These are relevantto any archaeological project, whether or not such a project is stimulated by a development proposal. The role ofarchaeomagnetic analysis at each stage is set out below.

PlanningThe integration of archaeomagnetic dating into an archaeological project must generally be undertaken in a morereactive way than is the case for mostother specialist services. This is because it is often not clear that suitable featureswill be uncovered on a site until excavationis at a relatively advanced stage. Evenwhen the presence of a substantialthermoremanent feature is anticipatedprior to excavation, the likelihood ofdisturbance since its remanence wasacquired cannot be assessed until it is exposed for inspection.

However, the requirement forarchaeomagnetic analysis should beconsidered at the planning stage of an archaeological project. Backgroundinformation concerning the site maysuggest whether suitable features are likely to occur – for instance, is industrialactivity involving kilns or furnaces expectedto be present? The potential precision of archaeomagnetic dating for the likelyage of the site, as compared with otherpossible dating techniques, should also be considered. However, where a suitable feature can already be welldated by other means, the Europeanarchaeomagnetic community would still be interested in sampling it as noted above in Future developments.The sections in Part 1,What can be dated?and Dating precision and limitations,provide guidance when assessing thesuitability of archaeomagnetic dating for a particular project. Further information

about how archaeomagnetism iscomplemented by other scientific datingtechniques can be found in Clark (1987)and Aitken (1990; 1999) and advice onthe application of the various physicaldating techniques is available from theEnglish Heritage Scientific Dating Team(see Appendix 1).

As the orientation of archaeomagneticsamples must be precise and the areas tobe sampled carefully chosen, a specialistwill normally be required to visit the siteduring excavation to undertake thesampling. It is important to bear in mindthat, at the time of writing, there are fewarchaeomagnetic practitioners who areable to provide a regular dating service in the UK. Since none of these specialistsis dedicated exclusively to providingarchaeomagnetic dates, their availabilityto work on a particular project may be limited. Therefore, such specialistsshould be contacted as soon as possible,preferably prior to the onset of excavation,for projects where archaeomagneticanalysis is likely to be required and they will need to discuss the:

● potential scale of the work;● types of features that are likely to need

sampling;● timescale for the excavation phase of

the project;● dates by when results will be required.

Such early discussion allows for scheduledsite visits, thus avoiding unnecessaryinterruptions and delays to tight excavationdeadlines, especially if it is envisaged that a large number of features will need sampling.

FieldworkThe above notwithstanding, it is often only during the fieldwork phase of a project that the requirement forarchaeomagnetic analysis is identified.If an initial survey component is involvedin the fieldwork, then information aboutthe possible presence of suitable featurescan be gained before excavation.Collection of surface finds can suggestthat industrial processes that are likely tohave involved kiln or furnace structurestook place at the site (English Heritage2001) and geophysical survey can detectthe presence of the remains of suchstructures (English Heritage 1995).Magnetic prospecting techniques areparticularly useful as they can oftendiscriminate anomalies possessing TRM.

SuitabilityWhen excavation commences, potentialfeatures can be inspected to confirm theirsuitability for archaeomagnetic dating.The section,What can be dated?, describesthe considerations involved and a briefchecklist is presented in Table 2.1.

With regard to thermoremanentmagnetisation of clays and clay soils,these often exhibit visible reddening whencompared to unfired samples of the samematerial. Unfortunately this visual test isnot always diagnostic as the precise colourchange will vary depending on mineralcomposition of the clay. Furthermore,Canti and Linford (2000) caution thatsoils in temperate northern hemisphereclimates may not exhibit significantreddening even when subjected to hightemperatures. In such cases, magneticsusceptibility measurements can be used as an additional tool to identify areas that might have been exposed to suitable temperatures in antiquity(Linford and Platzman 2004; Linford and Welch 2004). Further advice on theapplication of archaeomagnetic analysis to specific archaeological features can be sought (see Appendix 1) and theprovision of photographs or plans of the feature(s) under consideration canassist this process if a site visit is notimmediately possible.

Once the suitability of a feature for analysis has been established,arrangements can be made for anarchaeomagnetic dating specialist to visit and collect samples. Prior to thevisit, features to be sampled shouldideally be kept covered to avoid eitherexcessive waterlogging, or drying andshrinkage of surfaces in intense sunlight.If weather conditions permit, the covercan be removed on the day of the visit to allow any condensation to dry from the surfaces to be sampled.

Safe working practiceAs with all work on archaeological sites,archaeomagnetic dating specialists shouldcarry out their work under a definedhealth and safety policy and observe safe working practices at all times.Risk assessments should be carried out and documented where necessary.On building sites and archaeologicalexcavations specialists must also complywith the health and safety policies of the contractor. For further informationsee SCAUM (Standing Conference ofArchaeological Unit Managers 1991).

14


SamplingThe sampling and orientation process is described in the Sampling proceduresection above. It should be noted thatwhilst modern magnetometers allow very small samples to be taken, a degreeof damage to the feature being sampled is still necessary. Depending upon thesampling technique employed and thetype of material, it may be necessary toalmost entirely remove a small feature.It is prudent, therefore, to ensure that any context to be sampled has been fully recorded prior to archaeomagneticsampling. Copies of any availablephotographs and plans can also be useful to the archaeomagnetic datingspecialist. The specialist should:

● make a sketch plan showing the positionsof the samples within the feature;

● note relevant site context and samplenumbers for future reference.

Care should be taken to avoid exposingsamples to strong magnetic fields prior to measurement.

Assessment of potential for analysisDuring an archaeological project, oncefieldwork is completed, the materialrecovered will be catalogued and assessed

to determine what potential, if any,it offers for post-excavation analysis.A report (in MAP2, the assessment report)will be produced detailing this information.Based upon this assessment, a decision will be taken as to whether post-excavationanalysis of the material is warranted. If so,then the original project design will needto be updated to take account of thearchaeological material actually recoveredand its potential for further study.A revised project design document willthen usually be produced (in MAP2, theupdated project design) making proposalsfor the analyses deemed appropriate andthe project steering group will need toagree the required costs and resourcesbefore post-excavation analysis proceeds.

As with the other archaeological materialrecovered during fieldwork, samples takenfor archaeomagnetic analysis will need tobe included in this process. Preliminaryarchaeomagnetic results can be producedwithin about four weeks, so dating workcan, in some cases, proceed interactivelywith excavation.This is generally possibleonly when a small number of features are to be dated and the dating specialistdoes not have a queue of analyses toprocess. However, in such cases, finalarchaeomagnetic dates and a final

archive report (see below) may already beavailable at the time when the potential of other archaeological material is beingassessed. Final archaeomagnetic resultscan then be included in the assessmentreport.

On larger projects, when such rapidturnaround of final results is not possible,initial measurements of the samples canusually still be provided at the assessmentstage.These measurements can indicate if a feature is likely to have been disturbedand is thus not archaeomagneticallydatable. With these results, the projectmanager and dating specialist can agreepriorities for analysis, to ensure that themost archaeologically important, andarchaeomagnetically most promising,features are dated first. If analysis of thearchaeomagnetic samples would also havethe potential to address any other researchissues, these should be identified, and the necessary work itemised, at this stage. This information can then beincluded in the assessment report and,if appropriate, a timetabled programme of archaeomagnetic analysis can be drawnup for the updated project design.The bulkof the archaeomagnetic analysis wouldthen be completed in parallel with otherpost-excavation archaeological analyses.

15

Table 2.1 Checklist of factors influencing the suitability of a feature for archaeomagnetic analysis

All features Is the feature still in the same position as it was when it acquired its remanence?

This is a fundamental requirement of the archaeodirectional method. The feature should be inspected for cracking that might indicate that it has moved or been disturbed since firing/deposition.

In the absence of cracking, has the feature slumped or lost its structural integrity?

If there is evidence for slumping (see Part 1,The mean remanent direction) can the feature be assumed to have been level originally?

Is it possible to estimate the direction and degree of movement (strike and dip)?

Is the feature free of bioturbation, eg tree roots, mole activity?

At a smaller scale, is the material comprising the feature still well consolidated? For instance, loose sand or soil may become friable and individual particles may have moved realigning their stored magnetic field directions.

Is the feature likely to have been in close proximity to ferrous material when it acquired its remanence,eg an iron-smelting furnace that cooled with slag left inside it?

This can distort the magnetic remanence recorded in the feature.

Thermoremanent Has a suitably intense firing event taken place? (Usually determined by changes in the coloration of the fabric or features magnetic susceptibility measurements.)

If the feature is composed of clay, has it been baked hard?

Depositional Is it likely that the sediment has settled out of solution in low-energy conditions so that a DRM can form?remanent features


Analysis and report preparationOnce the updated project design has beenaccepted, any additional archaeomagneticanalysis required to produce dates for thesampled features can proceed. This willinclude the measurement of the naturalremanent magnetisation (NRM) of any samples not yet analysed and theproduction of demagnetisation curvesfor all, or a representative subset of, thespecimens (Clark et al 1988). If detaileddemagnetisation curves are not measuredon all the specimens, then those remainingshould be demagnetised using the set of optimal partial demagnetisationincrements identified from the detailedcurves. The direction of the primarymagnetisation component recorded ineach specimen can then be accuratelydetermined using principal componentsanalysis (Kirshvinck 1980). Statisticalanalysis of the final results for all thespecimens can then be carried out toproduce a mean direction of remanentmagnetisation for the feature which can be compared with the relevant calibrationcurve. More information about theseprocesses and references will be found inPart 1 under Laboratory measurement and Dating precision and limitations.

The archaeomagnetic dating reportOnce the results of the archaeomagneticanalysis have been completed they shouldbe presented as an archive report that isintelligible to the layperson as well as thespecialist; this report may be edited forinclusion in the final project report and its suggested form and content areoutlined below.

The primary function of the report is to record any dates deduced by thearchaeomagnetic analysis but it is alsoessential to present the results in such a way that they could be replicated byanother archaeomagnetic specialist.Regardless of how the archaeologicalinvestigation was instigated, anarchaeomagnetic dating report isnecessary to provide sufficient data to support conclusions drawn from the archaeomagnetic analysis.

The report should include pertinentinformation to allow for:

● assessment of the continuing validity of the archaeomagnetic date in the lightof future advances;

● future recalibration of a date as the UK archaeomagnetic calibration curveis refined;

● an archaeomagnetic date that hassupporting independent dating evidenceto be incorporated into the UKArchaeomagnetic Database as part of the ongoing improvement of UK calibration data.

The minimum requirements necessary for a report to meet these criteria are:

1 The identification and description ofthe feature including cross referencesto any context or feature numbersassigned by the excavator.

2 The location of the feature should be stated in terms of its latitude andlongitude to an accuracy of at least 0.1 degrees (the direction and magnitude of the Earth’s magnetic field varies withposition on the Earth’s surface).

3 The number and composition of samplestaken from the feature should be stated,as well as whether they were subsequentlydivided into specimens in the laboratory.If samples were subdivided, the numberof specimens derived from each sampleshould be recorded.

4 The partial demagnetisation regimeapplied to the specimens to removesecondary magnetisation componentsshould be described. Details of theprincipal components analysis for each specimen should be listed, eg thecalculated MAD (maximum angulardeviation) angles (see Kirshvinck 1980).

5 The number of samples, N, used tocalculate the feature’s mean remanentmagnetisation should be stated. Ifsamples were subdivided into specimensthen the method used to determine themean magnetisation vector for eachsample should be described. It is alsoessential to outline the number ofspecimens and/or samples rejectedfrom these calculations and the reasons for their exclusion.

6 The α95 circle of confidence angle for the calculation should be stated,in addition to the declination andinclination of the mean remanentdirection. It is also desirable to statethe associated estimate of the Fisherindex (K). A precision of one decimalplace is usually sufficient for all thesevalues, provided that angles aremeasured in degrees. See Tarling(1983, 117–22) for details of thecalculation of precision parameters for Fisher distributions.

7 The mean magnetisation direction foreach of the features analysed should be quoted before any corrections oradjustments are applied.

8 It should also be stated if a magneticdistortion correction was applied to the mean remanent direction beforecomparison with calibration data (Aitkenand Hawley 1971; Shuey et al 1970).

9 The adjusted values of the declinationand inclination should also be stated if the mean remanent direction wasadjusted to a central reference locationsuch as Meriden for comparison withcalibration data. If calibration was viathe calculation of a VGP then theinferred pole position and the polarerror parameters (Tarling 1983, 127)should be provided instead.

10 It is desirable to discuss, alongside thearchaeomagnetic date in the report,any other available chronologicalevidence that can bracket the date ofthe feature and which is independentof the archaeomagnetic analysis.

The exact layout of an archaeomagneticdating report will necessarily varydepending on such factors as the specificmethodologies employed and the numberand type of features sampled. However,most reports can conform to an outlinemodel, as suggested in Table 2.2 opposite,which allows all the specific pieces ofinformation listed above to be included in the relevant places.

DisseminationArchive reports for archaeomagneticanalyses funded by English Heritage aresubmitted for inclusion in the relevantEnglish Heritage report series. Thesereports make available the results ofspecialist investigations in advance of fullpublication and are available on requestfrom English Heritage (see Appendix 1).

The archaeomagnetic results should also beincluded in the report publication. By thetime the analysis and report preparationstage of the project is complete (see theprevious section) the project team will havedecided which of the following options topursue with regard to the presentation ofthe archaeomagnetic analysis:

● a summary of the archaeomagneticresults included in the main report text,while the archaeomagnetic report andrelated data is retained in archive;

● a summary of the results included in themain report text while the archaeomagneticreport is included as an appendix;

● the archaeomagnetic report is reproducedin the main report text (with only thosemodifications necessary for the sake of consistency).

16


17

Table 2.2 Outline model of an archaeomagnetic dating report

function key features

Summary identifies the dating study, its objectives and results

Introduction outlines the background to the project, ● the sampling location(s) (especially latitude and longitude) and date(s)personnel involved, dates of sampling and ● identification of the sampled features (context, feature and/or sample the location of the site numbers)

Methodology describes the techniques and instrumentation ● a description of the sampling and orientation strategies employedused to sample and analyse the feature ● the number and composition of the samples recovered and any (text should be supported by annotated subsequent division into individual specimensfeature plans and/or photographs of the ● the techniques and equipment employed for laboratory measurement feature(s) sampled showing sample positions, of the remanent magnetisation. (Where standardised procedures are see Figures below) used, this description is often placed in an appendix and only departures

from the standard methodology are noted here.)

Results discusses the results of the above ● the partial demagnetisation applied to each sample and discussion of measurements (text should be supported the reasons why any samples were rejectedby tables listing the results of all ● the calculated mean remanent direction for each feature analysed in measurements and interpretive graphical terms of declination and inclination (and intensity if determined) along illustrations, see Tables below) with the number of samples used and the associated precision parameters

● any corrections made to the mean remanence direction to compensate for magnetic distortion (unless already noted under Methodology)

● any adjustment to a standard reference location required to allow the calculated mean to be compared with calibration data (the mean parameters both before and after adjustment should be stated)

● a comparison of the mean remanent direction with calibration data and,if it is possible to date the feature, the derived date range

Conclusion summarises the main findings of the ● if an archaeomagnetic date could not be obtained for a feature possible archaeomagnetic analysis reasons should be discussed

● if a date was obtained, the date range can be compared with any other independent dating evidence. Such discussion is particularly important where crossovers or contiguities in the archaeomagnetic calibration curve result in two or more alternative possible date ranges.

Acknowledgements

Date summary summarises the minimal information ● reiterates information that should have already been noted in the text.requirements outlined above for each date However, it can be useful to summarise key points for ease of reference,

for example:laboratory feature code 1DFarchaeological feature identification brick kiln, context 1034location longitude 0.9°W, latitude 51.3°Nnumber of samples/specimens 16/16number of samples used in mean 16AF demagnetisation applied 10–100mTmean MAD angle of samples used in mean 0.5°distortion correction applied nonedeclination (at Meriden) -7.4° (-7.4°)inclination (at Meriden) 74.4° (75.1°)α95 0.9°k 1762.3date range (63% confidence) AD 1700 to 1720 date range (95% confidence) AD 1690 to 1725 archaeological date range AD 1698 to 1721 (documentary)

Tables should include ● NRM measurements of magnetisation declination, inclination and intensityfor all specimens/samples and the same measurements after optimalpartial demagnetisation

● measurements of declination, inclination and intensity at each partialdemagnetisation stage for all samples for which full demagnetisation curves were measured

Figures should include ● annotated feature plans and/or photographs of the feature(s) sampled showing sample positions

● appropriate plots of the distribution of sample magnetisation directions both for their NRM directions and after optimal partial demagnetisation

● plots of the demagnetisation curve(s) for a representative subset of those samples for which full demagnetisation curves were measured

● graphs comparing the mean remanent direction of each feature dated with the relevant calibration information

(often placedat the end ofthe report toaid readability)

(often placedat the end ofthe report toaid readability)


It should be noted that under theCopyright, Designs and Patents Act 1988the organisation or person undertakingfield and reporting work retains thecopyright to the material, unless this hasbeen varied in the contract for the work.This position should be made clear to all relevant parties at the outset of work(Institute of Field Archaeologists 1999,appendix 6).

As for all project work, close liaisonbetween the archaeomagnetic datingspecialist and the project team is essential – no less so at this stage where it will be important to ensure that thearchaeomagnetic analysis is presented inappropriate proportion to its contributionto the stated objectives of the project.

Summarising archaeomagnetic dates in publicationsWhere the results of archaeomagneticanalysis are to be summarised in a projectpublication, a certain minimum amount of information should be included foreach feature dated to allow it to beevaluated and compared with otherarchaeomagnetic results. Where possiblethe archaeomagnetic summary sectionshould include all of the minimallyrequired information listed in the section on the archaeomagnetic dating report.However, when considerations of spacemean that this is not possible, thefollowing parameters must be quotedalong with the date range for eacharchaeomagnetic date:

● the declination and inclination of the mean remanent direction at the site specified in degrees to one decimal place;

● the number of samples used to calculatethe mean direction, N;

● the α95 precision parameter in degrees to one decimal place. Preferably also the estimate of the Fisher index (K)quoted to unit value.

It is presumed that the location of the site,and the features dated, can be determinedfrom information elsewhere in the report.If this is not the case then the longitudeand latitude of the features dated shouldalso be stated in degrees to at least onedecimal place.

A draft of any written work that includesarchaeomagnetic results should always be sent back to the archaeomagneticspecialist for checking. This will avoid any misrepresentation of the results.

Citing archaeomagnetic datesArchaeomagnetic dates should be citedusing the formula: <oldest date> to<youngest date> at <confidence level>%confidence. For example, AD 1465 to1510 at 95% confidence or 100 BC to AD 30 at 95% confidence. Recentarchaeomagnetic analyses will provide adate range at the 95% confidence leveland, where available, it is this that shouldbe quoted. However, a date range at adifferent confidence level may be all thatis available with dates produced in thepast. In all cases the discussion shouldprovide an indication of where the moredetailed information outlined above may be found.

Data archivingIt is essential that a report on anyarchaeological intervention, even if it goesno further than an evaluation, should belodged as promptly as possible with theHistoric Environment Record (HER), amainly local authority-based service.This is necessary to inform future interventionsand guide the local planning authority onfuture decisions. The archaeomagneticanalysis should form part of this overallproject report. However, if no otherarchaeological investigation takes place as part of the project, a copy of thearchaeomagnetic dating report shouldinstead be sent to the HER.

The National Geophysical Data Center in Boulder, Colorado, USA houses theInternational Archaeomagnetic Databasefrom which the UK calibration data aremainly derived (Tarling and Dobson 1995).This database contains archaeomagneticanalyses of features from many parts of the world together with informationabout any independent evidence that can establish the date of the feature.A sizeable subset of the data is fromthe UK and has been used to compile theUK archaeomagnetic dating calibrationcurve. The database is available athttp://www.ngdc.noaa.gov/seg/geomag/paleo.shtml

Archaeomagnetic dating reference curves have largely been built up from an accumulation of analyses of featuresthat can also be dated using otherindependent evidence. Hence, anysuccessful archaeomagnetic analysis where some complementary datingevidence is also available should be considered for inclusion in theInternational Archaeomagnetic Database by the specialist concerned.

Case studiesThe following case studies describe twosituations where archaeomagnetic datinghas proved an effective aid to datingarchaeological features.They are intendedto illustrate how the principles andtechniques described in this document are applied in practice and, whererelevant, cross-references are made to the appropriate sections in Part 1.

Dogmersfield House brick kiln(thermoremanent magnetisation)In 2003, during a watching brief in advance of construction work at Dogmersfield House near Fleet,Hampshire, archaeologists from WessexArchaeology discovered the remains of an updraught brick kiln (context 1034 in Wessex Archaeology 2003).The St Johnfamily inherited Dogmersfield House in 1712 and the present house was builton the site in 1728, possibly incorporatingan earlier Elizabethan house. Furthersubstantial extension works were made in 1744 and it was thought that the kilnrelated to one of these two constructionphases. It was brick-built and alignednorth-east to south-west, consisting of two flues with a large stokepit at thesouth-western end (Fig 15). Subsequentexcavation showed that it had been builton top of the remains of a smallerupdraught kiln on a different alignmentand it appears that the later kiln wasconstructed as soon as the earlier one was abandoned, possibly because thelatter could not manufacture bricks insufficient quantities. The survival of the remains of two updraught kilns isrelatively rare in the Hampshire regionwhere most of the remains so far studiedhave dated from the 19th century andbeen of the more substantial downdraughtdesign (Moore 1988). Given theimportance of the discovery and itspotential to relate to historical evidence as well as the likelihood that fired bricksfrom the kiln would contain a TRM, theEnglish Heritage Geophysics Team wasasked to sample it for archaeomagneticanalysis.

Seven bricks were removed from the kilnbut, before disturbing them, orientationdiscs were attached using the disc method(see above Sampling procedure) andoriented relative to true north using agyro-theodolite. In the laboratory, 16specimens were obtained by cutting awayapproximately 8cc of the brick materialimmediately beneath selected discs (somediscs were left attached to the remaining

18


Table 2.3 NRM measurements of specimens and principal components of measurementsafter partial AF demagnetisation for the Dogmersfield brick kiln

NRM measurements after partial demagnetisationsample deco inco J(mAm-1) deco inco MAD angleo R

1DF01 -14.6 77.2 1698.6 -8.1 75.9 0.5

1DF02 -20.0 74.7 27999.4 -20.3 74.5 0.4

1DF05 -8.6 77.6 6453.4 -5.2 74.3 0.7

1DF06 -17.8 75.1 7848.2 -15.0 73.7 0.6

1DF07 -13.0 70.2 5059.6 -7.9 72.2 0.4

1DF08 -6.3 76.4 1980.4 -6.6 76.5 0.3

1DF10 4.8 77.5 21366.0 -7.0 74.5 0.4

1DF11 5.7 74.3 24532.8 -4.7 73.1 0.5

1DF12 23.0 76.4 6180.3 5.7 74.3 0.7

1DF13 -11.3 75.1 989.6 -10.3 75.4 0.3

1DF14 -13.9 73.8 1873.9 -8.9 73.9 0.4

1DF16 0.3 74.0 5924.0 1.1 74.5 0.5

1DF17 -4.9 74.3 12101.2 -8.4 74.5 0.5

1DF18 -12.8 73.1 3526.8 -9.5 73.9 0.6

1DF20 -2.0 74.3 11518.2 -2.9 75.3 0.6

1DF21 -15.0 72.2 8454.6 -9.8 73.0 0.6

NotesJ = magnitude of magnetisation vectorMAD = maximum angular deviation (see text and Kirshvinck 1980)R = sample rejected from mean calculation

material for future analysis).The specimenswere then partially demagnetised (seeabove Laboratory measurement) using a series of alternating field strengthsranging between 1mT and 100mT.Their directions of magnetisation were re-measured after each demagnetisationincrement and principal componentsanalysis was then used to examine thelinearity of the demagnetisation curvesproduced for each specimen. Inspection of the MAD angles indicated that all wereacceptably linear between the 10mT and100mT demagnetisation steps, someviscous overprinting being evident in thedirections measured before the 10mTdemagnetisation increment. As a rule ofthumb, MAD angles less than ~2° indicateacceptable linearity and an acceptablylinear demagnetisation curve suggests thatthe magnetisation is likely to be stable.The optimal direction of magnetisationrecorded by each specimen was calculatedusing the measurements in this range andthese are listed in Table 2.3.

A mean TRM direction was calculated (see above The mean remanent direction)from the primary TRM directionsdetermined for all 16 specimens usingstandard Fisher statistics. No correctionfor magnetic refraction was made as nosystematic variation in the magnetisationdirections of samples was observed in relation to their position within the kiln. The calculated mean TRMdirection was:

at sitedec = -7.4° inc = 74.4°α95 = 0.88° K = 1762.3

at Meriden dec = -7.4° inc = 75.1°

In the directions quoted above a negative sign for the magnetic declinationindicates a direction to the west of truenorth. This archaeomagnetic directionwas measured at Dogmersfield, Hampshire(longitude 0.9°W, latitude 51.3°N),so it was converted to the equivalentdirection that would have pertained at Meriden (longitude 1.6°W, latitude52.4°N) to allow for comparison with UK calibration data. The converteddirection was compared with a modifiedversion of the calibration curve of Batt(1997) (see above Dating precision andlimitations) which incorporated additionalcalibration data from the UK and northernFrance and which has been assembled since Batt’s curve was produced.

19

Fig 15 The Dogmersfield House brick kiln during sampling, looking south.


Fig 16 depicts the resulting calibrationcurve graphically with the mean TRMdirection from the Dogmersfield kilnsuperposed upon it along with the latter’s95% confidence limits. The date rangededuced for the last firing of the kiln isAD 1690 to 1725 at the 95% confidencelevel. This is earlier than either of the two documented construction events but accords remarkably well withinformation discovered subsequentlywhich indicates that the Dogmersfieldestate leased a house to John Reading,a bricklayer, between 1698 and 1719.Hence it is likely that the kiln was inoperation earlier than first thought andprobably supplied bricks for alterations or repairs to the house and garden wallsimmediately after the St John familyinherited the estate.

Yarnton ditch sediment (post-depositional remanentmagnetisation)Archaeological excavations in the 1990s byOxford Archaeology in advance of gravelextraction at the village of Yarnton, nearOxford revealed a wealth of settlementactivity on the floodplain of the northbank of the Thames. The underlyingarchaeological landscape consisted of a concentration of prehistoric funeraryactivity, focused around a Neolithicenclosure as well as evidence for domesticsettlement. A ditch (feature 12036,pictured in Fig 8) located ~600m to thewest of the Neolithic enclosure appearedto mark a boundary between the ritualelement of the site and the occupationactivity found elsewhere on the floodplain.Excavation of the ditch in 1997 suggestedthat the sediment comprising its fill waslikely to have been deposited from a bodyof water and tests by the English HeritageGeophysics Team indicated that itpossessed significantly enhanced magneticproperties compared to the surroundingsubstrate. An organic-rich layer wasidentified at the base of the ditch and themost strongly magnetised sediment wasassociated with it. Subsequently sufficientmaterial for radiocarbon dating was alsoextracted from this layer. Given theimportance of this boundary for theinterpretation of the site and the likelihoodthat a depositional or post-depositionalremanence was recorded by the sediment,it was sampled for archaeomagnetic datingwhilst the ditch section was exposed.

Sampling was carried out by pushingcylindrical 10cc plastic pots into the faceof the ditch section and marking arrows

20

a)

b)

c)

80o

70o

60o

50o

0o10o

30 o

20 o-10o

-20o

-30o

900

1000

1100

1200

13001400

1500

1600

1700

1900

1800

Declination

Inclination

800-30

-20

-10

10

20

30

0

1000 1200 1400 1600 1800 2000

Date (AD)

800 1000 1200 1400 1600 1800 2000

Date (AD)

Dec

linat

iono

50

55

60

70

75

80

65

Incl

inat

iono

Fig 16 Comparison of the mean remanent direction recorded by the Dogmersfield brick kiln with the archaeomagneticdating calibration curve of Batt (1997). (a) Equal angle stereogram showing the variation of declination and inclination withtime (dates are all AD, negative declinations are west of true north). (b) Variation of declination with time showing error barsdetermined for control points. (c) Variation of inclination with time, showing error bars.The mean thermoremanent directioncalculated for the kiln with its 95% confidence limits has been superimposed on all three diagrams.


on the base of each indicating the directionof vertical (see above Sampling procedure).A gyro-theodolite was used to establishthe vertical orientation as well as the angleof the face of the ditch section relative totrue north. Thirty-one sediment sampleswere recovered from the section (sample26 failed during extraction) by excavatinga small area around each pot to extract itthen sealing the open ends with airtight lids.No subdivision of the samples was carriedout in the laboratory so in this case eachsample taken from the site equated to anarchaeomagnetic specimen. The NRM of each specimen was measured beforeany partial demagnetisation (see aboveLaboratory measurement) was carried outand this revealed that those taken fromthe top of the ditch section (specimens 01 to 11) contained very little magneticmaterial. The remaining specimensderived from parts of the section either in or near the organic-rich layer and these were partially demagnetised using a series of alternating field strengthsranging between 1mT and 199mT.Their directions of magnetisation were re-measured after each demagnetisationincrement and principal componentsanalysis was then used to examine thelinearity of the demagnetisation curvesproduced for each specimen. All themeasured magnetisation directions arelisted in Table 2.4.

As greater measurement error is oftenencountered when analysing weaklymagnetised sediments, a MAD angle ofless than, or equal to, 3° was consideredto indicate acceptable linearity whenexamining the specimens’ demagnetisationcurves. Using this criterion it was foundthat specimens 15 and 22 did not retainstable magnetisation directions and thesesamples were excluded from the calculationof the mean remanence direction. Specimen21 was also excluded as its remanencedirection was anomalously far from thecluster of directions formed by the otherspecimens. A check on the site notesrevealed that this specimen, along withspecimen 22, had been located next to astone inclusion within the sediment andthis may well have distorted the recordedmagnetisation direction. The optimaldirections of magnetisation determined for the remaining 17 specimens wereaveraged using standard Fisher statistics(see above The mean remanent direction).No correction for magnetic refractionwas made as no systematic variation in the magnetisation directions of samples was observed in relation to

21

Table 2.4 NRM measurements of specimens and primary magnetisation components afterpartial AF demagnetisation for ditch 12036 at Yarnton, Oxfordshire

NRM measurements after partial demagnetisationsample deco inco J(mAm-1) deco inco MAD angleo R

01 0.9 75.4 0.6335 - - - R

02 3.4 75.7 0.6346 - - - R

03 79.3 63.9 0.4900 - - - R

04 -0.4 46.5 0.6228 - - - R

05 -88.8 -19.3 6.5037 - - - R

06 7.9 54.6 0.3636 - - - R

07 4.5 68.5 0.7046 - - - R

08 8.6 68.0 1.6987 - - - R

09 0.3 64.1 9.5434 - - - R

10 19 66.6 2.2329 - - - R

11 32.4 81.0 0.5625 - - - R

12 3.9 74.8 21.0829 4.5 75.1 1.7

13 25.4 76.6 37.9892 18.2 78.1 2.7

14 12.2 70.1 7.6265 14.5 69.7 2.0

15 3.9 77.2 10.1836 4.9 77.7 4.4 R

16 14 68.8 36.5633 16.9 68.7 1.1

17 30.2 61.3 8.3905 23.6 69.6 3.0

18 20.3 73.4 46.8495 15.3 74.4 2.7

19 16.1 68.2 58.1766 11.1 68.9 2.7

20 -4.5 63.6 19.1388 -3.0 64.0 1.7

21 -3.5 72.7 205.3024 32.4 65.0 1.7 R

22 16.9 69.6 68.0523 18.9 67.8 7.5 R

23 6.8 66.2 83.7510 8.4 66.1 2.6

24 8.6 69.5 161.7178 -15.4 70.5 2.1

25 10.7 66.8 59.3086 3.4 68.3 2.5

27 -12 69.3 43.9266 -11.2 69.8 1.9

28 13.9 74.8 37.7221 -8.9 76.2 2.4

29 0.3 65.3 78.5048 3.8 64.4 2.5

30 -5.2 70.7 96.0492 -2.8 67.6 3.0

31 7.8 69.9 82.3529 0.9 70.8 2.1

32 -15.6 69.6 33.1742 -1.6 64.6 1.0

NotesJ = magnitude of magnetisation vectorMAD = maximum angular deviation (see text and Kirshvinck 1980)R = sample rejected from mean calculation


their position within the ditch section.The calculated mean remanence direction was:

at sitedec = -4.2° inc = 70.1°α95 = 2.4° K = 214.7

at Meriden dec = -4.3° inc = 70.5°

In the directions quoted above the positivesign for the magnetic declination indicatesa direction to the east of true north. Thisarchaeomagnetic direction was measuredat Yarnton, Oxfordshire (longitude 1.3°W,latitude 51.8°N), so it was converted tothe equivalent direction that would havepertained at Meriden (longitude 1.6°W,latitude 52.4°N) to allow for comparisonwith UK calibration data. The converteddirection was compared with the calibrationcurve of Clark, Tarling and Noël (1988) as the prehistoric portion of this curve isbased on the largest amount of calibrationdata (see above Dating precision andlimitation). Fig 17 depicts the calibrationcurve graphically with the mean remanencedirection from the Yarnton ditch sedimentsuperposed upon it along with the latter’s95% confidence limits. The date rangededuced for the locking-in of the sedimentis 215 BC to 85 BC or 15 BC to AD 90at the 95% confidence level. This date is in good agreement with radiocarbondeterminations from two macro-fossilsamples of aquatic plant material recovered

22

-30 -20 -10 0 10 20 30 40 50

50

60

70

80

Declinationo

Incl

inat

iono

200300

400

500100

100BC

0

200BC300BC 500BC 600BC

800BC1000BC400BC

Fig 17 (above) Comparison of the mean remanent direction recorded by the Yarnton ditch sedimentwith the archaeomagnetic dating calibration curve of Clark,Tarling and Noël (1988) using a Bauer plot.The mean remanent direction together with its 95% confidence limits have been superimposed on the curve as a cross and shaded ellipse. Dates with no suffix are AD, negative declinations are west of true north.

Fig 18 (right) Transmission electron micrograph of the magnetic particles extracted from thesediments extracted from the ditch at Yarnton.The distinctive size range, morphology and absence of metal impurities all suggest that these magnetite particles were created by magnetotactic bacteria.

from the organic layer which producedcalibrated dates of 360 to 1 cal yr BC(OxA-10707) and 390 to 90 cal yr BC(OxA-10708) respectively. Furthermore,the radiocarbon dates suggest that theearlier of the two archaeomagnetic dateranges is to be preferred. Additionalmagnetic analysis suggested that theremanence carrier was likely to be almostpure magnetite (Linford et al 2005) and transmission electron micrographs of magnetic particles extracted from the sediment indicated morphologiescharacteristic of biogenic magnetite(magnetosomes) derived from magnetotacticbacteria (Fig 18). These bacteria indicatethat a micro-aerobic environment existedin the ditch in the late Iron Age and theirnumbers would have rapidly increasedwhilst conditions within the sediment layerremained favourable. As each generationof bacteria died their magnetosomes wouldhave become aligned with the ambientdirection of the Earth’s magnetic field,forming the post-depositional remanentmagnetisation measured in the samples.


Glossary

alpha-95 (αα95, circle of confidence)a measure of the uncertainty associatedwith a mean three-dimensional directionvector as calculated using Fisherianstatistics. The mean direction isimagined to lie along the axis ofrotational symmetry of a cone, directedaway from the point towards the circularbase. The α95 statistic is defined as thesemi-angle of this cone such that thereis a 95% probability that the true meandirection is fully contained within it (see Fig 13c and Aitken 1990, ch 9).The smaller the α95 angle, the smallerthe circular base of the cone (hence theterm circle of confidence) and thus themore precisely the mean direction hasbeen determined. As a very approximaterule of thumb, for archaeomagneticpurposes, an α95 greater than 5° wouldgenerally be considered poor precisionwhilst a value less than 2.5° typicallyindicates good precision.

angle of dip see inclinationanisotropy (magnetic anisotropy,

anisotropy of susceptibility) the ease with which a magnetic materialbecomes magnetised can depend on thedirection of the magnetising field.Thiscan lead to the direction of remanentmagnetisation acquired by the materialdiffering from that of the applied field(aligning more closely with an easydirection of magnetisation). Anisotropy is typically caused by either crystallinealignments in samples containinghaematite or shape alignments insamples containing magnetite.

antiferromagnetic (antiferromagnetism)a form of ferromagnetism where thecrystalline material contains two equallybut oppositely magnetised sublattices.If the directions of magnetisation of the two lattices are exactly opposite,zero net magnetisation will result.However, if the sublattice magnetisationdirections are not exactly opposed(canted antiferromagnetism) they willnot entirely cancel each other and aweak net magnetisation can result atapproximately 90° to the sublatticemagnetisation directions. Haematite (α-Fe2O3) exhibits this form ofpermanent magnetisation.

archaeodirection the direction of the geomagnetic field in antiquity. Thisterm is generally used to describe thearchaeodirectional method of magneticdating where the direction (rather thanthe intensity) of the geomagnetic field(and thus the apparent location of the

magnetic poles) is determined for thetime at which an object acquired its remanence.

archaeointensity the strength or intensity of the geomagnetic field inantiquity.This term is generally used todescribe the method of magnetic datingwhere the strength (rather than thedirection) of the geomagnetic field isdetermined for the time that an objectacquired its remanence.

blocking temperature (unblocking temperature) associated withthermoremanent magnetisation(TRM). On the cooling of a substancecontaining magnetic minerals, this is the temperature at which TRM becomes‘frozen in’.This temperature will dependon the precise mineralogical compositionof the substance as well as its crystallineorganisation (eg large or small grains).Materials with blocking temperaturesbelow about 200–250°C often do notretain fully stable magnetisations atroom temperature and may exhibitviscous remanence.When consideringthe heating of materials to remove theirremanence, it may be referred to asunblocking temperature.

coercivity the strength of magnetic field that must be applied to an objectpossessing remanent magnetisation, inthe opposite direction to its direction of magnetisation, to reduce its externalmagnetic field to zero whilst the coercivefield remains switched on. Coercivity isoften used in the analysis of the magneticmineralogy of samples.

coercivity of remanence (Bcr or Hcr) the strength of magnetic field that must be applied to a material exhibitingremanent magnetisation, in the oppositedirection to its direction of magnetisation,to result in the removal of that remanence(demagnetisation) after the coercivefield is switched off. Alternating fielddemagnetisation is often used to studythe coercivity of remanence of materialsin archaeomagnetic studies.

Curie temperature (Curie point) on heating, the temperature above which a material loses its ferrimagneticproperties. The blocking temperature of a particular mineral is related to itsCurie temperature but may be lowerowing to such considerations as chemicalimpurities, crystal size and shape.Named after Pierre Curie (1859–1906).

declination the angle in the horizontal plane between magnetic and true north.

Directions to the east of true north are considered positive, those to the west negative

depositional (detrital) remanent magnetisation (DRM) a remanentmagnetisation acquired as a sedimentis deposited. In archaeomagnetic terms,this is usually due to particles ofsediment rotating to align their intrinsicmagnetisations with the ambient field as they settle out of a relatively non-turbulent water solution. They thenbecome locked into position by theweight of sediment settling above them.

easy axes one or more directions, relativeto the crystal lattice, along which it isenergetically favourable for a crystal tobecome magnetised. The crystal willalways magnetise parallel to one ofthese directions which are referred to as its easy axes. The energy required to make the magnetisation direction flip to an alternate easy axis within the crystal will generally depend on the precise composition, shape and size of the particular grain.

ferrimagnetic (ferrimagnetism)a form of ferromagnetism where the crystalline material contains twooppositely but unequally magnetisedsublattices resulting in a net overallmagnetisation. For example magnetite(Fe3O4) where the iron atoms form asublattice magnetised in one direction,opposed by the sublattice of oxygenatoms that are magnetised in theopposite direction.

ferromagnetic (ferromagnetism) permanent magnetisation resulting fromstrong exchange interactions betweenneighbouring atoms in a crystal lattice.This causes the magnetisations of allatoms in the lattice to spontaneously alignin the same direction. Iron (Fe) is themost common ferromagnetic substance.In a looser sense ferromagnetism can be used as an umbrella termencompassing all related forms ofpermanent magnetisation such asantiferromagnetism andferrimagnetism.

Fisher (Fisherian) statistics a system of statistics developed to characterisethe distribution of unit vectors in three-dimensional space (Fisherian statisticsdescribes the variation of directionswhilst normal, Gaussian, statisticsdescribes the variation of scalarquantities). Developed by RonaldAylmer Fisher (1890–1962) (Fisher 1953).

23


fluxgate device designed to measure the strength of the ambient magnetic fieldalong a single axis. It consists of two,oppositely wound solenoids with coresconsisting of a soft magnetic materialthat reach their saturation magnetisationin a weak magnetic field.These solenoidsare both magnetised using the samealternating electric current and in the absence of any external field,the magnetic fields they generate willcancel each other out. However, in thepresence of a constant ambient fieldthere will be a phase difference betweenthe times at which the oppositelymagnetised cores are saturated resultingin a measurable alternating magneticfield. The strength of the ambientmagnetic field in the direction of thelong axes of the solenoids can beinferred from the amplitude andfrequency of this alternating field.

geomagnetic field the Earth’s spontaneously generated magnetic field. Largely due to movements ofelectrically conductive material in theEarth’s molten outer core but with asmaller magnitude contribution fromionic movements in the upperatmosphere.

geomagnetic reversal the phenomenon observed in the magnetisation ofancient rocks whereby the direction ofthe geomagnetic field appears to haveperiodically reversed (ie the magneticnorth pole exchanges position with themagnetic south pole). The period oftime between reversals is known as a chron and, in the recent geologicalpast, these seem to have lasted justunder one million years.We are presentlyin the Brunhes chron, named afterBernard Brunhes (1867–1910), and the last geomagnetic reversal (Brunhes-Matuyama) occurred about 780,000years ago.

grain a macroscopic sample of a crystalline mineral will generally consistof multiple conjoined crystals of varyingshapes and sizes. Each of these crystals,within which the atoms are usuallyarranged on a single regular lattice,is termed a grain.

gyro-theodolite a device capable of finding the direction of true north usingthe precession of a built-in gyroscope.

inclination (angle of dip) the angle between the local geomagnetic fielddirection and the horizontal plane.Conventionally downward directionsare considered positive. In the present

(Brunhes) chron, inclinations are typicallypositive in the northern hemisphere andnegative in the southern hemisphere.The magnetic poles are defined as thoseplaces on the Earth’s surface where theinclination angle is vertical (downwardsat the north pole, upwards at the south).

magnetic domain a region within a crystal of a magnetic mineral withinwhich the magnetisations of all atomsare parallel. For certain minerals andcrystal sizes, this may result in eachentire grain being magnetised in thesame direction (single domain behaviour),whilst in other cases, a grain might bedivided into several magnetic domainseach magnetised in a different direction(multi-domain behaviour).

magnetic moment a measure of the strength of a magnetic dipole measuredin Ampere metres squared (Am2) in the SI system

magnetic refraction the phenomenon by which the direction of an ambientmagnetic field changes as it crosses theinterface between materials with differentmagnetic properties, analogous to theway that light rays refract when crossingthe boundary between materials withdifferent refractive indices. It has beenpostulated that this effect may beresponsible for anomalous distortions to the directions of the magnetisationobserved in different parts of kilnstructures. However, the changes inmagnetisation observed in practice donot always match those predicted by thetheory, suggesting that other, less wellunderstood, factors are also involved.

natural remanent magnetisation (NRM) the remanence of a naturalsample as first measured in the laboratory(before any partial demagnetisation).The term implies nothing about theorigin of the remanence which could be thermoremanence, depositionalremanence, etc.

partial demagnetisation the process of removing the less stable components of a sample’s magnetisation (resultingfrom magnetic domains with lowblocking temperatures) to isolate thehigh coercivity or blocking temperaturecomponent. May be achieved viaexposure to an alternating magneticfield (AF demagnetisation), heating in an oven to a specific temperature(thermal demagnetisation) or byirradiation with low-power microwaveradiation (microwave demagnetisation).

post-depositional remanent magnetisation (pDRM) a remanentmagnetisation acquired after a sedimentis deposited. This can occur when adepositional remanent magnetisationdoes not become locked in until sometime after the initial deposition of the sediment or when the initialdepositional magnetisation is modifiedby chemical or other effects.

precision parameter (Fisher index)a dimensionless measure of the relativescatter of directions in a Fisherianstatistical distribution. Its value canrange from zero (directions drawn fromthe distribution are completely randomand uncorrelated) to infinity (completealignment on a single direction).

remanent magnetisation (magnetic remanence) permanent magnetisationof an object or material that persistseven after the removal of anymagnetising field.

secular variation gradual changes in the strength and direction of the geomagneticfield over time. Archaeomagnetic datingis only possible because of this temporalvariation in the Earth’s magnetic field.

specimen in archaeomagnetic studies a sample of magnetised material takenfrom a feature may be divided into a number of smaller specimens formeasurement. Measurements from all specimens from the same sample are usually averaged to determine asample mean, so reducing randommeasurement errors and differencescaused by material inhomogeneity.Hence, ‘specimen’ defines the unit of magnetic material upon whichmeasurements are conducted in thelaboratory whilst ‘sample’ refers to theunits of material originally extractedfrom the feature.

spinner magnetometer laboratory magnetometer capable of measuringweak magnetisations within samples.It exploits the fact that, when thesample is rotated relative to a fixedcollecting coil (or ring fluxgate), thesample’s rotating magnetic field willgenerate an electric current inproportion to its strength.

SQUID magnetometer SQUID = superconducting quantum interferencedevice. A cryogenic laboratorymagnetometer that exploits thephenomenon by which magnetic fluxcan only take fixed discrete valueswithin superconducting materials.Particularly useful for measuring

24


extremely weak magnetic fields and for large samples sets (as theinstrumentation is amenable to semi-automation).

thermoremanent magnetisation (TRM)a remanent magnetisation acquired after a substance has been heated thencooled in an ambient magnetic field.

viscous remanent magnetisation (VRM) magnetic domains withblocking temperatures lower thanabout 200°C can realign theirmagnetisation directions even at room temperature if given enough time. Timescales for the process range from minutes to tens or even hundredsof years. Hence a magnetic materialcontaining a reasonable proportion ofsuch domains can exhibit a magnetisationthat slowly changes over time, trackingchanges in the geomagnetic field, albeitwith some lag.

virtual geomagnetic pole (VGP) at any point on the Earth’s surface thegeomagnetic field has a direction whichis conventionally expressed in terms ofits declination and inclination.TheVGP position is defined as that positionon the Earth’s surface where the north(or south) magnetic pole would have tobe situated to cause the observed fielddirection, assuming that the Earth’smagnetic field can be modelled as ageocentric dipole. Note that the truegeomagnetic poles will generally notexactly coincide with such inferredVGPs owing to local perturbations in the Earth’s magnetic field (ie thegeocentric dipole model is only a first-order approximation to the truegeomagnetic field).

25


Appendix 1

Contact addresses for UK archaeomagnetism

English HeritageWithin English Heritage the first point ofcontact for general archaeological scienceenquiries should be the regional EnglishHeritage advisor for archaeological sciencewho can provide independent non-commercial advice. Such advisors arebased either in universities or in theEnglish Heritage regional offices. Pleasecontact regional advisors currently basedin universities at their university address,using the regional office address as afurther contact point when necessary.

East of England (Bedfordshire,Cambridgeshire, Essex, Hertfordshire,Norfolk, Suffolk)Dr Jen HeathcoteEnglish Heritage regional office24 Brooklands AveCambridge CB2 2BUTel: 01223 582700Mobile: 07979 206699E-mail:[email protected]

East Midlands (Derbyshire,Leicestershire, Rutland, Lincolnshire,Nottinghamshire, Northamptonshire)Dr Jim WilliamsEnglish Heritage regional office44 DerngateNorthampton NN1 1UHTel: 01604 735 400E-mail:[email protected]

LondonDr Jane SidellUniversity College LondonInstitute of Archaeology31–34 Gordon SquareLondon WC1H 0PYTel: 0207 679 4928Fax: 0207 383 2572E-mail: [email protected]

English Heritage regional office23 Savile RowLondon W1S 2ETTel: 0207 973 3000

Note From July 2006 the London regionaloffice will be at1 Waterhouse Square138–142 HolbornLondon EC1N 2TQ

North East (Northumberland, CountyDurham, Tyne & Wear, Tees Valley)Mrs Jacqui HuntleyDepartment of ArchaeologyUniversity of DurhamScience LaboratoriesDurham DH1 3LETel and Fax: 0191 334 1137E-mail: [email protected]

English Heritage regional officeBessie Surtees House41–44 SandhillNewcastle upon Tyne NE1 3JFTel: 0191 269 1200

North West (Cheshire, Greater Manchester,Merseyside, Lancashire, Cumbria)Dr Sue StallibrassUniversity of LiverpoolDepartment of Archaeology, Classics and Egyptology (SACE)Hartley BuildingBrownlow StreetLiverpool L69 3GSTel: 0151 794 5046Fax: 0151 794 5057E-mail: [email protected]

English Heritage regional officeCanada house3 Chepstow StreetManchester M1 5FWTel: 0161 242 1400

South East (Kent, Surrey, East Sussex,West Sussex, Berkshire, Buckinghamshire,Oxfordshire, Hampshire, Isle of Wight)Dr Dominique de MoulinsInstitute of ArchaeologyUniversity College LondonRoom 204A31–34 Gordon SquareLondon WC1H 0PYTel: 0207 679 1539Fax: 0207 383 2572E-mail: [email protected]

English Heritage regional officeEastgate Court195–205 High StreetGuildford GU1 3EHTel: 01483 252000

South West (Cornwall, Isles of Scilly,Devon, Somerset, Dorset, Wiltshire,Gloucestershire, Bath and NE Somerset,Bristol, South Gloucestershire,North Somerset)Ms Vanessa StrakerEnglish Heritage regional office29 Queen SquareBristol BS1 4NDTel: 0117 975 0700 Email:[email protected]

West Midlands (Herefordshire,Worcestershire, Shropshire, Staffordshire,West Midlands, Warwickshire)Ms Lisa MoffettEnglish Heritage regional office112 Colmore RowBirmingham B3 3AGTel: 0121 625 6820Mobile: 07769 960022Email:[email protected]

Yorkshire Region (York, North Yorkshire,South Yorkshire,West Yorkshire, East Riding,North and North East Lincolnshire)Mr Ian PanterEnglish Heritage37 Tanner RowYork YO1 6WPTel: 01904 601 983Fax: 01904 601 999Mobile: 07967 706869E-mail:[email protected]

Specific advice on scientific dating andarchaeomagnetism in particular can besought from Alex Bayliss, the EnglishHeritage Scientific Dating Coordinator,and Paul Linford of the English HeritageGeophysics Team respectively. Advice isavailable to all, free of charge. A limitedarchaeomagnetic assessment and analysisservice is also available for features fromEnglish Heritage funded projects whereprior arrangements have been made.Archaeomagnetic dating requests forexcavations funded from other sourceswill also be considered, subject to theapproval of the relevant English HeritageInspector of Ancient Monuments, wherefeatures are of major archaeologicalsignificance or where good independentcomplementary dating evidence is available.

26


Scientific dating coordinatorAlex BaylissEnglish Heritage23 Savile RowLondon W1S 2ETTel: 0207 973 3299Email:[email protected]

Archaeomagnetic dating specialistPaul LinfordEnglish Heritage Geophysics TeamFort CumberlandFort Cumberland RoadEastneyPortsmouth PO4 9LDTel: 023 9285 6749Email:[email protected]

Other sources of adviceAdvice on aspects of archaeomagnetismmay be sought from the contacts below,as noted.

Dr Cathy BattUniversity of BradfordDepartment of Archaeological SciencesBradford BD7 1DPTel: 01274 233 533Fax: 01274 235 190Email: [email protected]

Dr Batt is the coordinator of theArchaeomagnetic Applications for theRescue of Cultural Heritage (AARCH)European research network (see Part 1,Future developments). She can provideadvice on all aspects of archaeomagneticdating. Research interests include calibrationof UK archaeomagnetic dates, dating inthe Northern and Western Isles of the UK and integration of archaeomagneticdating with other methodologies. Limitedcontract dating services can be provided.

Dr Mark W HounslowCentre for Environmental Magnetism andPalaeomagnetism (CEMP)Geography DepartmentFaculty of Science and TechnologyLancaster UniversityBailriggLancasterLA1 4YBTel: 01524 594 588Fax: 01524 847099E-mail: [email protected]

The CEMP performs contractarchaeomagnetic dating services and candeal with all types of archaeomagneticmaterials from clay linings, to moresubstantial heated wall rocks and bricks of any consistency and hardness, andstrength of magnetisation. A wide variety of environmental magnetic measurementsfor ‘ground-truthing’ of magnetic surveydata are also undertaken.

Prof Mark NoëlGeoQuest AssociatesRocksideDreemskerryMaugholdIsle of Man IM7 1BLTel: 01624 819364E-mail: [email protected]

GeoQuest Associates provide acommercial archaeomagnetic datingservice for material sampled throughoutthe UK and Ireland. Advice is freely givenby Prof Noël who has been researching in this area since 1973. Further interestsinclude the archaeomagnetic properties ofcave deposits and sediments, and variousnon-chronometric applications of thearchaeomagnetic method.

Chris ThomasMuseum of London Archaeology ServiceMortimer Wheeler House46 Eagle Wharf RoadLondon N1 7EDTel: 020 7410 2261Fax: 020 7410 2201E-mail: [email protected]

The Museum of London ArchaeologyService Geomatics Team provide contractarchaeomagnetic dating advice, assessment,and sampling service, with final sampleanalysis and report preparation carriedout by the GeoQuest archaeomagneticlaboratory. The Museum of LondonArchaeology Service operates across the UK and abroad.

27


Appendix 2

Brief notes on the history ofarchaeomagnetism

Archaeomagnetism, the study of themagnetisation of archaeological materials,has a long history to which these briefnotes cannot hope to do full justice.The earliest steps in the story involve the discovery of the principles upon which archaeomagnetism depends: themagnetic field of the Earth and remanentmagnetisation. More detailed overviews of these aspects are provided by Tarling(1983, 1–8) and Good (1988 – see theentries on geomagnetism and palaeo-magnetism) and Malin and Bullard(1981) provide a thorough catalogue of early observations of declination andinclination made in England includingshort biographies of many of theindividuals concerned.

The phenomena of magnetic attractionand repulsion appear to have been knownto the ancient Egyptians and Greeks. Theability of magnetised needles to point inspecific directions is first recorded by theChinese in the 1st century AD (althoughit had almost certainly been known for at least 300 years previously) and theChinese were also aware of declination byAD 720, noting that the compass did notpoint due south (south being the Chineseprime meridian). However, this knowledgedoes not appear to have reached Europeuntil much later and the first Europeanrecord of the magnetic compass is notuntil the 12th century AD. Europeanobservations of declination are notreported until the 16th century AD when the Flemish cartographer GerhardMercator and the Portuguese explorerJoão de Castro both reported differencesbetween magnetic direction and truenorth. However, some 15th-centurysundials and road maps are thought to have markings indicating compasssettings that differ from true north.The angle of dip or inclination was also first recorded in the 16th century by Hartmann (1544) and independentlyby Norman in 1576 (Norman 1581).However, the crucial breakthrough in terms of archaeomagnetic dating was the discovery that declination andinclination changed over time (secularvariation) and this was first realised by Henry Gellibrand in 1635 (Gellibrand 1635).

With respect to the acquisition of remanentmagnetisation, Delesse (1849) and Melloni(1853) were the first to discover thatvolcanic rocks were magnetised and bothconcluded that the magnetisation wasacquired on cooling. However, muchearlier Boyle (1691) had observed thatbricks became magnetised along their longaxes on cooling in the Earth’s magneticfield. Mercanton (1918) studied themagnetisation of ancient (2000-year-old)vase bases and showed that the individualmagnetisation directions were randomised.This indicated that they retained theiroriginal magnetisation acquired duringfiring rather than a common magnetisationdirection acquired in later fields afterdeposition in archaeological deposits.

The development of a scientific datingmethod based upon the above discoveries is a relatively recent innovation of the mid-20th century, mainly following fromThellier’s ground-breaking work in France(Thellier 1938). In the UK the firstarchaeomagnetic studies were by Cookand Belshé (1958). Subsequent analysis of lake sediment data (Thompson andTurner 1979; Turner and Thompson1981), well-dated archaeologicalstructures (Aitken and Hawley 1966;Aitken 1970; Clark et al 1988; Tarling and Dobson 1995), as well as collation ofdirect laboratory measurements over thelast 400 years (Malin and Bullard 1981),has led to the construction of the UKarchaeomagnetic calibration curve (Clarket al 1988; Tarling and Dobson 1995; Batt1997). In Europe, studies have also beenmade in Belgium (Hus and Geeraerts1998), Bulgaria (Kovacheva 1997; Lanoset al 1999), Denmark (Abrahamsen 1973,France (Bucur 1994; Lanos et al 1999),Germany (Schnepp and Pucher 2000;Schnepp et al 2004), Greece (Kovachevaet al 2000) and Hungary (Márton 1996;2003). Extensive investigation has alsobeen carried out in the American Southwest(Sternberg 1989; Eighmy 1991; Lengyeland Eighmy 2002).

Bibliography

Abrahamsen, N 1973 ‘Archaeomagnetictilt correction on bricks’. Archaeometry,15, 267–74

Aitken, M J 1970 ‘Dating by archaeo-magnetic and thermoluminescent methods’.Philosophical Transactions of the RoyalSociety of London A, 269, 77–88

Aitken, M J 1990 Science-based Dating in Archaeology. London: Longman

Aitken, M J 1999 ‘Archaeological datingusing physical phenomena’. Reports onProgress in Physics, 62, 1333–76

Aitken, M J and Hawley, H N 1966‘Magnetic dating – III: furtherarchaeomagnetic measurements inBritain’. Archaeometry, 9, 187–97

Aitken, M J and Hawley, H N 1971‘Archaeomagnetism: evidence forrefraction in kiln structures’.Archaeometry, 13, 83–5

Batt, C M 1997 ‘The Britisharchaeomagnetic calibration curve: anobjective treatment’. Archaeometry, 39,153–68

Batt, C M 1999 ‘Preliminary investigationsinto the acquisition of remanence inarchaeological sediments’ in Tarling, D Hand Turner, P (eds), Palaeomagnetism andDiagenesis in Sediments (Geological SocietySpecial Publication 151). London:Geological Society, 9–19

Boyle, R 1691 Experimenta et ObservationesPhysicae. London

Bucur, I 1994 ‘The direction of theterrestrial magnetic field in France, duringthe last 21 centuries. Recent progress’.Physics of the Earth and Planetary Interiors,87, 95–109

Canti, M G and Linford, N T 2000 ‘The effects of fire on archaeological soilsand sediments: temperature and colourrelationships’. Proceedings of the PrehistoricSociety, 66, 385–95

Casas, L, Shaw, J, Gich, M and Share, J A 2005 ‘High-quality microwavearchaeointensity determinations from anearly 18th century AD English brick kiln.’Geophysical Journal International, 161,653–61.

28


Clark, A J 1987 Scientific Dating Techniques.(IFA Technical Paper 5). Birmingham:Institute of Field Archaeologists

Clark, A J 1990 Seeing Beneath the Soil:Prospecting Methods in Archaeology.London: Batsford

Clark, A J,Tarling, D H and Noël, M 1988‘Developments in archaeomagnetic datingin Britain’. Journal of Archaeological Science,15, 645–67

Coey, J M D, Buochez, R and Dang, N V1979 ‘Ancient techniques’. Journal ofApplied Physics, 50 (B11), 7772–7

Collinson, D W 1983 Methods in RockMagnetism and Palaeomagnetism:Techniquesand Instrumentation. London: Chapmanand Hall

Cook, R M and Belshé, J C 1958‘Archaeomagnetism: a preliminary report on Britain’. Antiquity, 32, 167–78

David, A and Linford, P 1999‘Palaeomagnetic measurements’ inRoberts, M B and Parfitt, S A (eds),A Middle Pleistocene Hominid Site atEartham Quarry, Boxgrove,West Sussex,UK (Archaeological Report 17). London:English Heritage, 297–9

Delesse, A 1849 ‘Sur le magnétismepolaire dans les mineraux et dans lesroches’. Annales de chimie et de physique,25, 194–209

Department of Environment 1990Archaeology and Planning (Planning PolicyGuidance Note 16). London, HMSO

Eighmy, J L 1991 ‘Archaeomagnetism:new data on the south-west U.S.A. mastervirtual geomagnetic-curve’. Archaeometry,33, 201–14

English Heritage 1991 (reprinted 1996)Management of Archaeological Projects(MAP 2). London: English Heritage

English Heritage 1995 Geophysical Survey in Archaeological Field Evaluation.(Research and Professional ServicesGuideline 1). London: English Heritage

English Heritage 2001 Archaeometallurgy,(Centre for Archaeology Guidelines).Swindon: English Heritage

English Heritage (forthcoming)Management of Research Projects in theHistoric Environment (MoRPHE)

Evans, M E and Heller, F 2003Environmental Magnetism: Principles andApplications of Enviromagnetics. San Diego:Academic Press

Fisher, R A 1953 ‘Dispersion on a sphere’.Proceedings of the Royal Society of London A,217, 295–305

Gaffney, C and Gater, J 2003 Revealing the Buried Past: Geophysics forArchaeologists. Stroud: Tempus

Gellibrand, H 1635 A DiscourseMathematical on the Variation of theMagnetic Needle. London: William Jones

Good, G A (ed) 1998 Sciences of the Earth:An Encyclopedia of Events, People, andPhenomena. New York: Garland

Gutin, J C 1995 ‘Remains in Spain nowreign as oldest Europeans’. Science, 26,754–5

Hartmann, G 1544 ‘Letter’. Published in Sz. Ges. Endkunde, 32, 112–36

Hathaway, J H and Krause, G J 1990 ‘The sun compass versus the magneticcompass in archaeomagentic samplecollection’ in Eighmy, J L and Sternberg, R S (eds), ArchaeomagenticDating. Tucson: University of ArizonaPress, 109–34

Hoffman, K A 1988 ‘Ancient magneticreversals: Clues to the geodynamo’.Scientific American, 285, 50–9

Hus, J and Geeraerts, R 1998 ‘The directionof the geomagnetic field in Belgium sinceRoman times and the reliability ofarchaeomagnetic dating’. Physics andChemistry of the Earth, 9–10, 997–1007

Institute of Field Archaeologists 1999Standard and Guidance for ArchaeologicalField Evaluations (revised edn). Manchester:Institute of Field Archaeologists

Kirshvinck, J L 1980 ‘The least-squaresline and plane and the analysis ofpaleomagnetic data’. Geophysical Journal ofthe Royal Astronomical Society, 62, 699–718

Korte, M and Holme, R 2003‘Regularisation of spherical cap harmonics’.Geophysical Journal International, 153,253–62

Kovacheva, M 1997 ‘Archaeomagneticdatabase from Bulgaria’. Physics of theEarth and Planetary Interiors, 102, 145–51

Kovacheva, M, Spartharas,V and Liritzis, I2000 ‘New archaeointensity results fromGreek materials’. Archaeometry, 42, 415–29

Lanos, P 2004 ‘Bayesian inference of calibration curves: application toarchaeomagnetism’ in Buck, C E andMillard, A R (eds), Tools for ConstructingChronologies: Crossing Disciplinary Boundaries(Lecture Notes in Statistics 177). Berlin:Springer, 43–82

Lanos, P, Kovacheva, M and Chauvin, A1999 ‘Archaeomagnetism, methodologyand applications: implementation andpractice of the archaeomagnetic methodin France and Bulgaria’. EuropeanArchaeology, 2, 365–92

Lengyel, S N and Eighmy, J L 2002 ‘A Revision of the U.S. SouthwestArchaeomagnetic Master Curve’. Journal of Archaeological Science, 29, 1423–34

Linford, N, Linford, P and Platzman,E 2005 ‘Dating environmental changeusing magnetic bacteria in archaeologicalsoils from the upper Thames Valley, UK’Journal of Archaeological Science, 32 (7),1037–43

Linford, N T and Platzman, E 2004‘Estimating the approximate firingtemperature of burnt archaeologicalsediments through an unmixing algorithmapplied to hysteresis data’. Physics of theEarth and Planetary Interiors, 147 (2–3),197–207

Linford, P 1990 ‘Archaeomagnetic dating:Burton Dassett, Warwickshire’. AncientMonuments Laboratory Report. 8/90

Linford, P and Welch, C M 2004‘Archaeomagnetic analysis of glassmakingsites at Bagot’s Park in Staffordshire,England’. Physics of the Earth andPlanetary Interiors, 147 (2–3), 209–21

29


Malin, S R C and Bullard, E C 1981 ‘The direction of the Earth’s magnetic fieldat London, 1570–1975’. PhilosophicalTransactions of the Royal Society of London A,299, 357–423

Márton, P 1996 ‘Archaeomagneticdirections: the Hungarian calibrationcurve’ in Morris, A and Tarling, D (eds),Palaeomagnetism and Tectonics of theMediterranean Region (Geological SocietySpecial Publication 105). London:Geological Society, 385–99

Márton, P 2003 ‘Recent achievements in archaeomagnetism in Hungary’.Geophysical Journal International, 153 (3),675–90

McDougall, I and Tarling, D H 1983 ‘The magnetic sourcing of obsidiansamples from Mediterranean and NearEastern sources’. Journal of ArchaeologicalScience, 10, 441–52

Melloni, M 1853 ‘Du magnétisme desroches’. Comptes Rendus de l’AcadémieScientifique de Paris, 37, 966–8

Meng, Z and Noël, M 1989‘Archaeomagnetic evidence for the ageand duration of firing of mediaeval hearthsfrom Coffee Yard,York’. GeophysicalJournal International, 97, 357–9

Mercanton, P L 1918 ‘Etat magnétique de quelques terres cuites pré-historiques’.Comptes rendus des séances de l’Académie des Sciences de Paris, 166, 681, 949

Merril, R T, McElhinny, M andMcFadden, P L 1996 The Magnetic Field ofthe Earth: Paleomagnetism, the Core and theDeep Mantle. San Diego: Academic Press.

Moore, P 1988 The Industrial Heritage of Hampshire and the Isle of Wight,Chichester: Phillimore

Néel, L 1955 ‘Some theoretical aspects of rock magnetism’. Advances in Physics,4, 191–243

Noël, M 1996 ‘Archaeomagnetic analysisof two kilns at Heybridge, Maldon, Essex’.GeoQuest Associates unpublished report

Noël, M and Batt, C M 1990 ‘A methodfor correcting geographically separatedremanence directions for the purpose ofarchaeomagnetic dating’. GeophysicalJournal International, 102, 753–6

Norman, R 1581 The New Attractive,1st edn. London: Jhon Kyngston forRichard Ballard

Partridge, T C, Shaw, J, Heslop, D andClarke, R J 1999 ‘The new hominidskeleton from Sterkfontein, South Africa:age and preliminary assessment’. Journalof Quaternary Science, 14, 293–8

Rasmussen, K L 2001 ‘FOCUS: Provenanceof ceramics revealed by magneticsusceptibility and thermoluminescence’.Journal of Archaeological Science, 28, 451–6

Schnepp, E and Pucher, R 2000 ‘The German archaeomagnetic secularvariation curve’. Geol. Carpathia, 51, 184

Schnepp, E, Pucher, R, Reinders, J,Hambach, U, Soffel, H and Hedley, I 2004‘A German catalogue of archaeomagneticdata’. Geophysical Journal International,157 (1), 64–78

Schurr, K, Becker, H and Soffel, H C1984 ‘Archaeomagnetic study of medievalfireplaces and ovens and the problem ofmagnetic refraction during the past 6000years’. Journal of Geophysics, 56, 1–8

Shaw, J, Walton, D,Yang, S, Rolph, T Cand Share, J A 1996 ‘Microwavearchaeointensities from Peruvian ceramics’.Geophysical Journal International, 124,241–4

Shaw, J,Yang, S, Rolph, T C and Sun,F Y 1999 ‘A comparison of archaeointensityresults from Chinese ceramics usingmicrowave and conventional Thellier’s and Shaw’s methods’. Geophysical JournalInternational, 136 (3), 714–8

Shuey, R T, Cole, E R and Mikulich, M J1970 ‘Geographic correction ofarchaeomagnetic data’. Journal ofGeomagnetism and Geoelectricity, 22,485–9.

Standing Conference of ArchaeologicalUnit Managers (SCAUM) (1991).Health and Safety in Field Archaeology.[Lancaster]

Sternberg, R S 1989 ‘Secular variation of archaeomagnetic direction in theAmerican southwest, A.D. 750–1425’.Journal of Geophysical Research, 94, 527–46

Tarling, D H 1983 Palaeomagnetism:Principles and Applications on Geology,Geophysics and Archaeology, London:Chapman and Hall

Tarling, D H and Dobson, M J 1995‘Archaeomagnetism: an error assessmentof fired material observations in theBritish directional database’. Journal of Geomagnetism and Geoelectricity, 47,5–18

Tarling, D H, Hammo, N B and Downey, W S 1986 ‘The scatter ofmagnetic directions in archaeomagneticstudies’. Geophysics, 51, 634–9

Tauxe, L 2002 Paleomagnetic principles and practice. Dordrecht: Kluwer Academic.

Thellier, E 1938 ‘Sur l’aimantation des terres cuites et ses applicationsgeophysiques’. Annales de l’Institut dePhysique du Globe, 16, 157–302

Thompson, R and Oldfield, F 1986Environmental Magnetism. London:Allen and Unwin.

Thompson, R and Turner, G M 1979‘British geomagnetic master curve 10,000–0 yr BP for dating Europeansediments’. Geophys. Res. Letters, 6, 249–52

Turner, G M and Thompson, R 1981‘Lake sediment record of the geomagneticsecular variation in Britain duringHolocene times’. Geophysical Journal of the Royal Astronomical Society, 65,703–25

Wessex Archaeology 2003 ‘DogmersfieldPark, Hampshire: Archaeological WatchingBrief (2001–2003)’. Wessex Archaeologyunpublished report 44874.01

Williams-Thorpe, O and Thorpe, R S1993 ‘Magnetic susceptibility used in non-destructive provenancing of Romangranite columns’. Archaeometry, 35 (2),185–95

30


Acknowledgements

In its finished form this document owes a considerable debt of gratitude to manyindividuals both within and outside English Heritage. Within English Heritage AlexBayliss, Dr Andrew David and Dr Neil Linford have provided support as well as usefulcomments and advice throughout the guidelines’ development process. Andrew and Neilare particularly thanked for reading through complete drafts more than once. Gratitudeis also warmly extended to the English Heritage Regional Science Advisors, Lisa Moffett,Dr Dominique de Moulins, Peter Murphy, Ian Panter, Jane Sidell, Vanessa Straker andDr Jim Williams for providing valuable comments on the first full draft that have helpedto improve the usefulness and clarity of the final text. This draft was also circulated tocolleagues outside English Heritage and warm expressions of thanks are extended to Dr Cathy Batt, Dr Merion Hill, Dr Mark Hounslow, Prof Mark Noël, Dr Peter Rauxlohand Prof Don Tarling for providing often extensive comments and suggested amendmentsthat have greatly improved the accuracy of the finished work.

Thanks for the provision of photographs used in this document are due to Andrew David(Figs 9 and 10), Neil Linford (Figs 5 and 8) and Chris Welch (Fig 7) of English Heritageas well as Mark Hounslow (Fig 18) of the Centre for Environmental Magnetism andPalaeomagnetism, University of Lancaster and Nick Palmer (Fig 6) of Warwick FieldArchaeology. All other illustrations and photographs were created by Paul Linford,English Heritage.

Thanks are also due to Prof John Shaw for providing useful background informationand Jennifer Hillam whose Dendrochronology Guidelines provide an excellent template forthe layout of scientific dating guideline documents.

The overall text was compiled and written by Paul Linford. Every attempt has beenmade to present an evenly balanced, up-to-date and, above all, a useful document.Imperfections will certainly remain as well as areas where professional opinion isdivided. Furthermore, simply for expediency, it has not been possible to incorporateevery suggestion made by those who commented on drafts of the text. Revision iscertain to be required in the light of ongoing research and thus any comments towardsfuture editions would be welcome (see address in Appendix 1).

As part of the consultation process, a draft version of this document was made available tomembers of the AARCH network through the site www.meteo.be/CPG/aarch.net/linford.pdf

An abridged version of Part 1 was published as Linford, P 2004 ‘ArchaeomagneticDating’, Physics Education, 39 (2), 145–54.

These guidelines draw largely on the experience of the former English Heritage Centrefor Archaeology and its predecessor for archaeological science, the Ancient MonumentsLaboratory.

31

Cover figures: Research students from The University of Bradford collectingarchaeomagnetic samples at Metchley,Birmingham

Published March 2006

Copyright © English Heritage 2006Edited and brought to press byJoan HodsdonEnglish Heritage PublishingDesigned by Creative ServicesProduced by English Heritage PublishingPrinted by Wyndeham Westway

Product code 51162


English Heritage is the Government’s statutory advisor on the historic environment.English Heritage provides expert advice to the Government about all matters relating to the historic environment and its conservation.

For further information (and copies of this leaflet, quoting the product code 51162),please contact:English HeritageCustomer Services DepartmentPO Box 569Swindon SN2 2YPTelephone: 0870 333 1181Fax: 01793 414926E-mail: [email protected]

0406 - Archaeomagnetic Dating 24/3/06 10:37 am Page cov2

archaeomagnetic dating.ps, page 1-33 @ normalize ( 0406...

Documents