Archaeological and archaeomagnetic dating at a site from the ager Tarraconensis

16 Pages·2013·3.36 MB·English

by  Marta Prevosti

JournalofArchaeologicalScience40(2013)2686e2701
ContentslistsavailableatSciVerseScienceDirect
Journal of Archaeological Science
journal homepage: http://www.elsevier.com/locate/jas
Archaeological and archaeomagnetic dating at a site from the ager Tarraconensis
(Tarragona, Spain): El Vila-sec Roman pottery
Marta Prevostia, Lluís Casasb,*, Josep Francesc Roig Pérezc, Boutheina Fouzaid, Aureli Álvarezb,
Àfrica Pitarche
aInstitutCatalàd’ArqueologiaClàssica,Plaçad’enRovellat,s/n,43003Tarragona,Catalonia,Spain
bUniversitatAutònomadeBarcelona,FacultatdeCiències,DepartamentdeGeologia,CampusdelaUAB,08193Bellaterra,Catalonia,Spain
cCODEXArqueologiaiPatrimoni,Pl.SantFructuós1,43002Tarragona,Catalonia,Spain
dUniversitédeTunisElManar,FacultédesSciences,DépartementdeGéologie,CampusUniversitaire,2092ManarII,Tunisia
eIBeAResearchGroup,DepartmentofAnalyticalChemistry,FacultyofScienceandTechnology,UniversityoftheBasqueCountry(UPV/EHU),BarrioSarrienas/n,48940Leioa,Bizkaia,
Spain
a r t i c l e i n f o a b s t r a c t
Articlehistory: AveryaccuratearchaeologicaldatingofaRomansiteinNESpain(ElVila-sec)wasmadebasedonthe
Received1August2012 typologyofpotteryartifacts.Threedifferentphaseswereidentifiedwithactivityrangingfromthemid-
Receivedinrevisedform 1stcenturyBCtotheearly-3rdcenturyAD.AnalysesofbricksfromkilnsatElVila-secproduceddataon
11January2013
theirstoredarchaeomagneticvector.Thesedatawerecomparedwiththesecularvariationcurveforthe
Accepted24January2013
IberianPeninsulaandtheSCHA.DIF.3Kregionalarchaeomagneticmodel.Both,thereferencecurveand
themodel,producedprobabilitydistributionsforthefinalperiodofusefortwokilnsfromthesecond
Keywords: archaeologicalphasethatwerenotusedduringthethirdphase.Ata95%confidencelevel,bothtime
Archaeomagnetism
distributions cover a wide chronological range including the presumed archaeological age. Both the
Ceramic
Typology Iberian secular variation curveand the SCHA.DIF.3K regional model proved to be suitable models for
Dating dating the site, although on their own they do not produce a single unambiguous solution. This
Geomagneticfieldmodelling archaeomagnetic approach could also be applied to neighbouring archaeological sites that have an
Archaeodirection imprecisearchaeologicalage.
Archaeointensity (cid:1)2013ElsevierLtd.Allrightsreserved.
1. Introduction the availability of experimental data close to the area and time
under study (Casas et al., 2008). Consequently, archaeomagnetic
Sincetheappearanceofarchaeomagnetismasadatingtoolfor datingofsitesfromareaswithalimitedrecordofwell-datedsites
archaeologicalsites(Aitken,1974),suchdatinghasbeenbasedon appearstobesomewhatimpreciseand,inparticularwithregardto
theuseofsecularvariationcurves(SVC)(e.g.LeGoffetal.,2002; archaeomagneticintensities,largediscrepanciescanbefoundbe-
Gómez-Paccardetal.,2006;Zananirietal.,2007).Thesereference tween model predictions and actual data determinations (Fouzai
curvesarearbitrarilycentredona givenlocation andinvolve the et al., 2012; Schnepp et al., 2009). In contrast, in countries with
relocationofdatafromsitesthatarebothwell-dated(tobuiltthe an extensive archaeomagnetic record, the models tend to agree
SVC)andpoorly-dated(toattemptarchaeomagneticdating)sites. withthemeasureddata(e.g.Catanzaritietal.,2012).Datingisthus
The relocation of archaeomagnetic data to a central location is possible and in some cases the obtained ages can constrain the
aprocedurethatinvolvesaninherenterror(CasasandIncoronato, presumed ones derived from archaeological evidence (Gómez-
2007). The current trend in archaeomagnetic dating tools is to PaccardandBeamud,2008).
develop geomagnetic models of the secular variation that can be InthispaperwepresenttheresultsfromexcavationsatElVila-
usedtoobtainanadhocSVCforthespecificsiteweareinterested Sec pottery (NE Spain). The site comprises twelve kilns and the
in.Despitethis,thereliabilityofthesemodelsdependslargelyon findsmadeinsidethemallowaverypreciseidentificationofwhen
theywereabandoned.Archaeomagneticdatinghasbeenattemp-
ted for some of these kilns to verify the accuracy of archae-
* Correspondingauthor.Tel.:þ34935868365;fax:þ34935811263. omagnetic dating tools in NE Spain and to provide new
E-mailaddress:[email protected](Ll.Casas). archaeomagneticdatafromwell-datedarchaeologicalfeatures.
0305-4403/$eseefrontmatter(cid:1)2013ElsevierLtd.Allrightsreserved.
http://dx.doi.org/10.1016/j.jas.2013.01.027
M.Prevostietal./JournalofArchaeologicalScience40(2013)2686e2701 2687
2. ElVila-secpottery
2.1. Archaeologicalframework
ElVila-secisanextraordinaryarchaeologicalsiteintheTarraco
ager,duetothenumberofexcavatedanddatablekilns,aswellas
thevarietyofpotteryproducedthere,includingDressel(Dr.)2e4
amphorae (Dressel, 1899a; Sciallano and Sibella, 1993; López
Mullor and Martín, 2008), Hispanic terra sigillata, thin-walled
wareandoillamps.ThecoastalstripoftheHispaniaTarraconensis
provincewasanareafamousforitswineinantiquity(LópezMullor
andAquilué,2008;PrevostiandMartíniOliveras,2009).Amphorae
for wine bottling were manufactured at several potteries which
also produced building materials, coarse ware, terra sigillata and
sometimesevenothertypesof fineware.Itisvitaltoknowhow
thesepotteriesevolvedinordertotracethedevelopmentofeco-
nomiclifeinRomantimesandparticularlyintheCampdeTarra-
gona area, where such workshops are frequent and of
aconsiderablesize(PrevostiandGuitartiDuran,2010,2011).
A well-checked final dating for the wine amphorae from Tar-
ragonaisofgreatarchaeologicalinterest,asitiscrucialtoknowthe
chronologicalperiodsofsuchanimportanteconomicactivityasthe
tradeinTarragonawineinRomantimes(JárregaandOtiña,2008;
Berni Millet, 2010, 2011; Járrega and Prevosti, 2011), which is
mentionedinclassicalsources(Prevosti,2009).
2.2. Thearchaeologicalsite,excavationsandarchaeologicaldating
The widening of the C-14 highway between the towns of
Alcover and Reus in 2006 and 2007 allowed the existence of
apottery(figlina)withatotalsurfaceareaofaround2300m2tobe
confirmedintheareaofVila-sec(nearAlcover).Roig(2009,2010)
documentedtwelvekilnsandnumberedthemfrom1to12(Fig.1),
anumberingweuseinthisarticle.ElVila-secisoneofthelargest
Roman potteries discovered in the province of Hispania Fig.1. GeneralplanofElVila-secpottery;thekilnsarenumberedasinRoig(2009).
Tarraconensis.
Roig’s (2007, 2008, 2009, 2010) archaelogical excavations hel-
pedtoidentifytwodifferentzoneswithpotterykilnsandtodis- Mayet 36 (period from Tiberius to Nero), Mayet 37 (period from
tinguishthreechronologicalphases.Inadditiontothetwelvekilns, TiberiustoNero),Mayet38(30e35ADto75e80AD)andLópez54
other parts of the sitewere also excavated, including six settling (20BCe50AD)(Mayet,1975;LópezMullor,1989),aswellasother
basins for the clay, a kneading trough, tegulae channels, a large unclassified shapes that are identifiable as beakers or bowls. An
storeandsomeareasusedtoprocess,shapeanddrypotteryitems. earlysmall-scaleproductionofDr.2e4amphoraeassociatedwith
The production fromthis industrial workshop consisted of table- Kiln7(Fig.1)wasalsoreported.Thefirstphasecanthereforebe
ware,cookingware,thin-walledware,architecturalpottery,doliae, dated to between the Augustan era and the middle of the 1st
Hispanicterrasigillata,probablylampsandespeciallyamphoraeof centuryAD.
theDr.2e4andlateDr.2e4types. Thesecondphaseischaracterizedbytheabandonmentofallthe
The first identified phase of activity at the pottery consists of precedingkilnsandtheconstructionofasecondgroupoffourkilns
seven kilns (1e7 in Fig. 1). No pottery associated with the con- builtinaline(Kilns9e12,Fig.1).Thesewerelocatedatthesouth-
struction of these kilns was found as the construction trenches eastern end of the site and there was also a series of structures
were dug directly into the geological substratum. However, their relatedtotheproductionandstorageoffinishedpieces.Thecon-
stratigraphicpositiondatesthemtopriortothesecondidentified structionofthekilnscanbedatedtothemiddleofthe1stcentury
phase(mid-1stcenturyAD). AD,basedonstratafromthedestructionofthepreviouskilnsand
The pottery produced during this first phase consisted of thelevellingofthegroundtoprepareitforthenewbuildingworks.
building materials (tegulae, imbrex and antefixes), pondera, coarse It is worth to mentioning the materials that characterize these
ware (especially plates, casseroles, bowls, bottles, jars, mortars, strata: i) Italic terra sigillata Conspectus (Consp.) 18.2 (15 BCe30
basinsandlids),lamps,Hispanicterrasigillataandespeciallythin- AD), Consp. 23.1 (25e75 AD), Consp. 33.1 (1e50 AD) and Consp.
walledware.TheHispanic terrasigilataceramicsincludes(Fig.2) 36.3.2(10e30AD)(Ettlingeretal.,1990)forms;ii)SouthGaulish
imitations of South Gaulish terra sigillata goblets of the forms terrasigillataDrag.15a1(1e60AD),Drag.15b1(60e120AD),Drag.
Dragendorff (Drag.) 24/25b and Drag. 27b (Dragendorff, 1895; 17b(25e60AD),Drag.18a(15e60AD),Drag.24/25b(40e70AD),
Ritterling, 1913; Oswald and Pryce, 1920; Mezquíriz, 1961, 1985; Drag.27b(40e80AD),Drag.33a1(20e60AD),Haltern5(10e50
Passelac and Vernhet, 1993; Roca and Fernández García, 2005; AD) and Ritterling 12 (40e70 AD) (Dragendorff,1895; Ritterling,
Genin,2007).Thethin-walledwareincludes(Fig.3)Mayet18(10 1913;OswaldandPryce,1920;PasselacandVernhet,1993;Genin,
BCto60e70AD),Mayet21(periodofAugustus),Mayet29(period 2007) forms; iii) African coarse ware Ostia II, 306 (14e117 AD)
ofTiberius),Mayet30(periodofTiberius),Mayet33(periodfrom (Bertietal.,1970;Hayes,1972)form;iv)thin-walledwareMayet21
Augustus to Claudius), Mayet 35 (period from Tiberius to Nero), (25BCe25AD),Mayet30(10e50AD),Mayet33(10e30AD),Mayet
2688 M.Prevostietal./JournalofArchaeologicalScience40(2013)2686e2701
Fig.2. HispanicterrasigillatamadeatElVila-secduringthefirstphase(1e3,Drag.27;4e7,Drag.24/25;8e11,moulds).TheinventorynumberscorrespondtoRoig(2010).Dueto
thehighdegreeoferosionoffragments5and8e11,thenegativesofdecorativemotifshavebeenhighlightedtomakethemeasiertoobserve.
35(15e60AD),Mayet37(25e60AD)andLópez54(20BCe50AD) of the site. These especially concern the construction of Kiln 8
(Mayet,1975;LópezMullor,1989)forms;v)Amphorae,particularly (Fig.1)andalterationscarriedoutonKilns11and12(Fig.1)from
fromsouthernHispania,i.e.Dr.7e11(25BCe100AD),Dr.20B(30e thesecondphase.Datingofthethirdphaseisbasedonasoillevel,
50 AD) and Haltern 70 (50 BCe75 AD) forms, and among those that corresponds to the backfill level of Stratigraphic Unit 3092
producedinTarragona,Oberaden74(1e30AD),Pascual1(50BCe (Roig, 2010), which is related to the construction of an area
50 AD) and the ever-present Dr. 2e4 (25 BCe300 AD) (Pascual, delimitedbyrowsoflateDr.2e4amphoraeplacedtopdownwards
1977; Sciallano and Sibella,1993; Raynaud,1993a; López Mullor (Fig. 5). These amphorae datefromthe 2nd centuryand the first
and Martín, 2008); vi) also of note is a volute lamp, type Dr. 9B halfof the3rdcenturyAD(Járregaand Otiña, 2008).Withinthis
(Claudian era) (Dressel, 1899b; Pavolini, 1987; Morillo Cerdán, level, a number of pottery items was recovered reinforcing the
1989);andlastlyadupondiusfromTiberianperiod(21/22e37AD). datingofthisphasetothe2ndcenturyAD:Hispanicterrasigillata
The pottery manufactured during this second phase is clearly type Drag.15/17 (30e300 AD), Drag.37a (70e300 AD) and Drag.
made up of Dr. 2e4 amphorae and, to a lesser extent, building 37b (70e100 AD) (Mezquíriz, 1961, 1985; Roca and Fernández
material (tegulae and imbrex) and coarse ware (storage pots, García,2005).
cookingpots,casserolesandsmalljars)(Fig.4). This third phase is associated with the production of late or
Thethirdphaseofactivityatthepotteryisindicatedbyaseriesof evolvedDr.2e4amphorae,althoughsomemanufactureofbuilding
smallalterationsandfunctionalchangesvisibleatdifferentpoints materials and coarse ware cannot be excluded. In fact, the kilns
M.Prevostietal./JournalofArchaeologicalScience40(2013)2686e2701 2689
Fig.3. Thin-walledwaremadeatElVila-secduringthefirstphase(1e4,Mayet37form;5e8,Mayet30form;9and10,Mayet35form;11e12,Mayet33form;13and14Mayet29
form;15,originalbowlfromElVila-sec;16originalbeakerfromElVila-sec;17,Mayet21form:18,López54form).TheinventorynumberscorrespondtoRoig(2010).
fromthisphaseareeithertoosmall(Kiln8)forfiringamphoraeor 300 AD) pots and the lid of an Ostia III, 332 type pot (69e450
theyhaveevidenceofdirectfiring(Kilns11and12)andtherefore AD) (Lamboglia, 1958; Berti et al., 1970; Hayes, 1972; Raynaud,
amphoraewouldhavebeenproducedinKilns9and10. 1993b,c); ii) Hispanic terra sigillata, with of particular note the
Thishypothesiscannotbeconfirmeduntilthecompletionofthe Drag.15/17(30e300AD),Drag.18(50e200AD),Drag.24/25(30e
pendingexcavations. 150AD),Drag.30(50e100AD),Drag.35(50e150AD)andDrag.36
Thepotterystructureswereinfilledandabandonedinthelate2nd (50e300AD),Drag.37a(70e300AD)andDrag.37b(70e100AD)
orearly3rdcenturiesAD.Thematerialsthatallowthedatingofthe (Mezquíriz,1961,1985; Roca and Fernández García, 2005) forms.
uppermoststrataare:i)AfricanRedSlip(ARS)wareterrasigillata Thiscollectionofpotterydatestheabandonmentofthestructures
chiara A: Lamboglia (Lamb.) 2a (100e175 AD) and Lamb. 4/36A totheendofthe2ndcenturyAD,althoughthefindofanextremely
(90e175 AD) forms and ARS coarse ware; among the finds are eroded Antoninian coin datedtothe 3rd century AD extends the
worthyofmentionare:Lamb.10A(100e450AD),Lamb.10B(69e durationofthefinalfillofthepotteryworkshopremainsintothe
250 AD), Ostia III, 267A (100e475 AD) and Ostia III, 324 (100e nextcentury.
2690 M.Prevostietal./JournalofArchaeologicalScience40(2013)2686e2701
Fig.4. PotterymadeatElVila-secduringthesecondphase(1and2,amphoraeDr.2e4withthestampITA;3and4,jars;5and6,pots;7e10,covers;11:casserole).Theinventory
numberscorrespondtoRoig(2010).
Insummary,wecanstatethatthefirstphase(includingKilns1e Channel-Kiln8wasbuilt.Allfivekilns(secondandthirdphasesof
7) can be dated to the time of Augustus and that they were occupation) weredefinitivelyfilled in during the late2nd - early
rendered inoperative by the building of the second occupation 3rdcenturiesAD.
phase(which includesKilns 9e12) fromthe mid-1stcentury AD.
ItshouldbenotedthatexcavationofKilns9and10couldnotbe 2.3. Detaileddescriptionofthesampledkilns
completed and their chronological evolution is therefore not
known.However,Kilns11and12werefullyexcavatedandthere- Unfortunately, the remaining kiln structures from the first
foreitisknownthattheywerealteredinthemid-2ndcenturyAD, buildingphasewerequitedamaged.Thewallsoftheircombustion
withinwhatwecall thethirdphase. Itwasalsoat this timethat chambers were eroded and crumbling and thus not easily
M.Prevostietal./JournalofArchaeologicalScience40(2013)2686e2701 2691
Fig.5. LateDr.2e4amphoraemadeatElVila-secduringthethirdphase.TheinventorynumberscorrespondtoRoig(2010).
samplableforarchaeomagneticpurposes.OnlyKiln7(Fig.6)was 1971e1972, 1985, 2007). Three loose and quite crumbly quad-
successfullysampled. rangular bricks from the base of the arches from Kiln 7 were
The structure of Kiln 7 was partly destroyed by later building removedanddrilledinthelaboratory,andfinallycuttoproduce23
phases,soitwasnotpossibletodelimitthepraefurnium,althoughit standard cylindrical (w2.5 cm diameter) specimens for archae-
isrecognizablethattheaccesstothisopeningwasviaitsnorthern ointensitydeterminations.
side. An axial wall that divides the chamberlengthwise intotwo Threeofthefourkilnsbuiltinthesecondbuildingphasewere
waspreservedwithinthefiringchamber.Theremainsoffivearches sampled.TheseareKilns9(Fig.7),11(Fig.8)and12(Fig.9).Kiln9is
werealsovisible;theysupportedtheclaypipegrid,whichwasnot circularwithadiameterofabout3m.Thereisaperimeterstone
preserved.ThiskilncanbeincludedwithinCuomodiCaprio’sType and mud wall and an inner latericium or brick wall of the firing
II/c (rectangular with a double passageway) (Cuomo di Caprio, chamber. As the excavation could not be completed, an accurate
2692 M.Prevostietal./JournalofArchaeologicalScience40(2013)2686e2701
Fig.6. Kiln7withindicationofthesampledbricks.ThiskilnproducedDressel2e4amphoraebetweentheAugustaneraandthemiddleofthe1stcenturyAD.Therectangular
structurewithadoublepassagewaycorrespondstoN.CuomodiCaprio’s(1971e1972,1985,2007)TypeII/c.
classificationcannotbemade,althoughitmaybeTypeI/b(circular wallandanaxialinnerwallwerereportedforboth.Theperimeter
with a clay pipe grid supported bya radial wall) from Cuomo di walls are made of mud bricks, bricks and fragments of tegulae
Caprio’s (1971e1972, 1985, 2007) classification. Kilns 11 and 12 boundwithmud,whereastheaxialwallsofthefiringchamberare
are rectangular; the preserved parts are those excavated in the madeofbricksboundwithmud.SimilarlytoKiln7,onthethree
substratumi.e.thefiringchamberandthepraefurnium.Aperimeter walls that delimit the firing chamberof these kilns there are the
Fig.7. Kiln9withindicationofthedrilledcores.Thiskilnhasaperimetermudstonewallandaninnerfiringchamberwallmadeoflatericiumorbrickmaterial.ItisprobablyanN.
CuomodiCaprio(1971e1972,1985,2007)TypeI/b.
M.Prevostietal./JournalofArchaeologicalScience40(2013)2686e2701 2693
Fig.8. Kiln11withindicationofthesampledareasandlaterestructuringfeatures.Thiskilnwasbuiltduringthesecondphaseandrebuiltinthemiddleofthe2ndcenturyAD.The
originalrectangularstructurewithadoublepassagewaycorrespondstoN.CuomodiCaprio’s(1971e1972,1985,2007)TypeII/c.
remains of the arches that supported the clay pipe grid, which was made smaller, using the firing chamber from the previous
againwasnotpreserved.InthecaseofKiln12,aventilationholeor phase.Attherear,theupperpartoftheaxialwallwasdismantled,
airfluewasreportedinthewesterncorner.AccordingtotheCuomo anewmud-brickwallonitsnorth-westernsidewasbuiltandthe
di Caprio’s (1971e1972,1985, 2007) classification both kilns also firingchamberfloorlevelwasraised(Fig.8).Theresultisasmaller
correspondtoTypeII/c(rectangularwithdoublepassageway).Kils kilnwithtwoopeningsonallsidesoftheearlieraxialwall.Kiln12
11and12wererestructuredduringthethirdbuildingphase.Kiln11 underwentasimilarreorganization(Fig.9):theeasterncornerof
Fig.9. Kiln12withindicationofthesampledareas,theventilationholeandlaterestructuringfeatures.LikeKiln11,itwasbuildduringthesecondphaseandrebuiltinthemiddle
ofthe2ndcenturyAD.TheoriginalrectangularstructurewithadoublepassagewaycorrespondsN.CuomodiCaprio’s(1971e1972,1985,2007)TypeII/c.
2694 M.Prevostietal./JournalofArchaeologicalScience40(2013)2686e2701
theperimeterwallofthepreviousfiringchamberwasdismantled, microwavearchaeointensitydeterminations(attheGeomagnetism
a new transverse wall was built to move the firing chamber for- Laboratory, University of Liverpool). Archaeodirectional analyses
ward,lengtheningtheaxialwallwithanextensionmadeofbricks consisted of stepwise demagnetization of the natural remanent
andmudbricksboundwithmud,andthefloorlevelofthechamber magnetization(NRM)andmeasurementofthemagnetizationleft
wasraised.Theresultwasthatthefiringwasconcentratedbetween after each step. Thermal demagnetization was performed in
the oldaxial wall andthe earlier north-easternwall whereasthe a Schoenstedt TSD-1 demagnetizer and magnetization measure-
gap between the earlier axial wall and the earlier south-western ments on a 2G Enterprises superconducting rock magnetometer.
wallwasabandonedandfilledwithearth.Withthecurrentavail- Results were presented as Zijderveld diagrams (Zijderveld,1967).
abledata,itisnotpossibletodeterminewhethertherestructured Characteristic remanent magnetization (ChRM) directions were
kilns(11and12)werestillusedtofirepotteryoranotherkindof calculated by principal component analyses (Kirschvink, 1980).
material. Threeanomalousdirectionswereobtained(oneperkiln),possibly
Orientedsampleswereretrievedfromtheinternalopuslateri- duetoan incorrect replacementofbrokencylindersduringsam-
ciumwallsofKilns9(Fig.7),11and12forarchaeodirectionalde- pling.These outliers wereremovedfromthe calculation ofmean
terminations. For Kilns 11 and 12, sampling was performed on directions. Mean directions for each kiln and merged mean di-
structures not fired during the third archaeological phase: the rections were computed following Fisher (1953) statistics; con-
abandoned firing chamber belowthe new floorlevel(Fig. 8) and centration parameter k and confidence factor a95 were also
the abandoned south-western passageway (Fig. 9). A portable computed.Themeandirectionwascomparedwithi)theSVCfor
electricaldrillwithawater-cooleddiamondbitwasusedfollowing theIberianPeninsula(Gómez-Paccardetal.,2006)andii)SCHA.-
thestandardpalaeomagneticsamplingprocedure;eachcorepro- DIF.3K model (Pavón-Carrasco et al., 2009) predictions using
ducedasinglespecimen.Theinsituazimuthanddipofthecores aMatlabdatingtooldevelopedbyPavón-Carrascoetal.(2011).The
were measured using a compass coupled to a core-orienting fix- Iberian SVC is defined at a central location (Madrid) and it was
ture.Due tothe crumblyconsistencyof thewalls, drilling had to computed by the hierarchical Bayesian method using archae-
proceeddelicatelytoavoidbreakingthecylinderbeforetakingits omagneticdirectionswithdatesrangingfrom775BCto1959AD.
orientation.Ifbroken,thecylinderwascarefullyplacedbackinits The SCHA.DIF.3K model was obtained by least-sums of absolute
holetomarkitsorientation.Fromthreerejected,misorientedcyl- deviation inversion of palaeomagnetic data using spherical har-
inders from Kiln 12 (the first sampled), small cylinders (5 mm monics (SCHA) and provides full geomagnetic field vector values
diameter, 3 mm length) were sub-sampled in the laboratory for overtheEuropeancontinentandneighbouringareasfrom1000BC
microwavearchaeointensitydeterminations.SamplesfromKilns11 to1900AD.
and 12 had the most constrained archaeological age as theycor- ArchaeointensityanalyseswereperformedaccordingtotheCoe
respondtostructuresfiredduringthesecondarchaeologicalphase variant of a Thellier-type experiment (Coe, 1967), the NRM was
(mid1stcenturyAD-mid2ndcenturyAD),butnotduringthethird measuredandgraduallyremovedandreplacedbyanewthermal
one. In contrast Kiln 9 could actually have continued to operate magnetization (TRM). This was achieved by heating the samples
duringthethirdarchaeologicalphase. alternativelyinazero(Z)anda50mTapplied(A)fieldinaMagnetic
Tosumup,62specimenswereobtainedfrom4kilnsattheVila- MeasurementsMMTD-80thermaldemagnetizer.Besidesthecon-
secRomanpotterywithaveryprecisearchaeologicalage.Table1 ventional Z/A steps, pTRM and pTRM tail checks (Riisager and
summarizes the details of all the sampled structures, their pre- Riisager, 2001) were performed to ensure the absence of alter-
sumedarchaeologicalagesandthenumberofsamplestaken. ationandmultidomainbehaviourwithinthemagneticremanence
carriers. The remanent magnetization measurements were also
performed on a 2G Enterprises superconducting rock magneto-
3. Experimentalmethodsanddataanalyses meter.Coolingratetestswereperformedat560(cid:2)Conallsamples
following the procedure described in Chauvin et al. (2000). The
Rock magnetic properties were measured at the Geo-
correction was applied only for samples with stable acquisition
magnetism Laboratory, University of Liverpool for core samples
fromKiln7andforseveralrejectedmisorientedcoresfromKiln capacity (r2 < 5% where r2 ¼ (TRM1 (cid:3) TRM3)/TRM1). Microwave
archaeointensity analyses followed a similar procedure, although
12.Plate-likesampleswerecutfromthecoresandtheirchangein
using microwaves instead of heat to directly generate magnons
susceptibilitywasmeasuredastheywarmedfromliquidnitrogen
(Shawetal.,1999)toerasetheNRMortogenerateanewmicro-
toroomtemperature.Hysteresisproperties,isothermalremanent
magnetization(IRM)acquisition,remanencecoercivityandmag- wave thermoremanent magnetization (TMRM). The microwave
systemusedoperatedinthe14GHzfrequencyrange;throughout
netization versus temperature curves were obtained by using
the experiments the frequency was finely tuned to the ferro-
a Magnetic Measurements variable field translation balance
magnetic resonant frequency of the sample and after irradiation
(VFTB).
allowed direct measurement of the magnetization. Both conven-
Three types of palaeomagnetic procedures wereapplied: con-
tionalandmicrowavearchaeointensityresultswererepresentedas
ventional archaeodirection and archaeointensity determinations
(atthePalaeomagneticLaboratoryofBarcelona,SCTUB-CSIC);and AraiplotswhereNRMlostisplottedagainstTRM(orTMRM)gained,
bothnormalizedtotheinitialNRMalongwiththepTRMandtail
checks (Yu and Dunlop, 2003). Additionally Zijderveld diagrams
Table1 werealsoplottedusingthestepsperformedinzerofieldtocheck
SampledfeaturesinElVila-sec(Latitude:41.24(cid:2)N,Longitude:1.17(cid:2)E)withindication
the directional uniformity of the NRM vector. An overall archae-
oftheirpresumedagesaccordingtoarchaeologicalevidence,thesamplelabels,the
numberofcollectedcores(N)andtheappliedarchaeomagnetictechnique. ointensityvaluewascomputedforeachkilnbyfittingaGaussian
function to the sum of all individual results (Fouzai et al., 2012).
Kiln Presumedage Label N Technique
Samples with negative pTRM checks or f values lower than 0.5
7 50BCe50AD K7 23a Conventionalarchaeointensity
9 50ADe225AD K9 5 Conventionalarchaeodirection (Biggin and Thomas, 2003) were not used to obtain the overall
11 50ADe150AD K11 9 Conventionalarchaeodirection intensity estimate. Positive pTRM checks were defined as those
12 50ADe150AD K12 9 Conventionalarchaeodirection withadifferencebetweentheoriginalpTRMandthepTRMcheck
12 50ADe150AD K12 3 Microwavearchaeointensity lower than 10 percent of the total TRM acquired (Chauvin et al.,
a From3quadrangularbricks. 2000).
M.Prevostietal./JournalofArchaeologicalScience40(2013)2686e2701 2695
Fig.10. RepresentativeZijderveldplotsdepictingtheorthogonalprojectionoftheremanentmagnetizationvectorsduringprogressivedemagnetizationforspecimensfromKiln9
(left),Kiln11(middle)andKiln12(right).Open(solid)symbolsrepresentprojectionsonvertical(horizontal)planes.LinesindicatetheChRMdirections.
Thefullarchaeomagneticvector(includingbothmeandirection aprocedurethatinvolvesaninherenterror(CasasandIncoronato,
and intensity) was also compared with the previously described 2007).However,theuseofarchaeomagneticfieldmodelsinvolves
SCHA.DIF.3K model. The use of an archaeomagnetic field model regularizationandsmoothingtointerpolatefieldvalues,whichcan
avoids the need for relocation of data to a central location, alsobeasourceoferror.
Fig.11. Stereographicprojectionofthearchaeomagneticdirectionscalculatedforeachsample.In(a)meandirectionsanda95errorcircleswerecalculatedandplottedseparatelyfor
eachkiln,in(b)theindividualresultsfromtheKilns11and12weremergedtoproduceasinglemeandirectionanda95error.Nindicatesthetotalnumberofsamplesfromthetwo
kilns;DandIstandsfordeclinationandinclination;a95andk,95%confidenceconeofthemeandirectionandprecisionparameterfromFisherstatistics.