Dendrochronological dating and provenancing of timbers from the Arade 1 shipwreck, Portugal

Dendrochronological Dating and Provenancing of Timbers fromthe Arade 1 Shipwreck, Portugal

Marta Domínguez-DelmásNetherlands Cultural Heritage Agency and Ring Foundation-Dutch Centre for Dendrochronology; Smallepad 5,3800 BP Amersfoort, The Netherlands

Nigel NaylingSchool of Archaeology History and Anthropology, University of Wales Trinity Saint David, Lampeter, Ceredigion,SA48 7ED, United Kingdom

Tomasz WaznyLaboratory of Tree-Ring Research, University of Arizona, 105 W Stadium, Tucson AZ 85721 USA and NicolausCopernicus University, Institute for the Study, Conservation and Restoration of Cultural Heritage, ul. Sienkiewicza30/32, 87-100 Torun, Poland

Vanessa LoureiroInstituto de Arqueologia e Paleociências das Universidades Nova de Lisboa e do Algarve, Avenida de Berna, 26-C,1069-061 Lisbon, Portugal

Catherine LavierLaboratoire d’Archéologie Moléculaire et Structurale, CNRS-UMR8220, Université Pierre et Marie Curie, Sited’Ivry – Le Raphaël, 3 rue Galilée, 94200 Ivry-sur-Seine, France

As part of a larger project promoting the development of historical dendrochronology in the Iberian Peninsula, ship-timbersfrom the Arade 1 wreck (mostly planking and framing elements), stored at the DANS/IGESPAR in Lisbon, were examined. Ofthese, 52 samples were identified as deciduous oak (Quercus subg. quercus) and two as chestnut (Castanea sativa). Of 24 timbersselected for dendrochronological research, 23 could be dated, placing the origin of the wood in western France and the fellingof trees between AD 1579 and 1583. Their homogeneity suggests they are part of the original construction, which probably tookplace shortly after AD 1583.

© 2012 The Authors

Key words: Iberia, France, oak, chestnut, wood identification, 16th-century shipbuilding.

The application of dendrochronological analysisto timbers from wooden shipwrecks has, incertain geographical regions, become the norm

in nautical archaeology. The potential of this scien-tific discipline to provide precise dates for the fellingof trees where the bark-edge (the last ring formedbefore felling) survives on some timbers makes den-drochronology attractive and particularly useful forcultural-heritage studies. Its high precision (down tothe year or even the cutting season) is possiblebecause the sequence of thinner and thicker rings on

a sample is characteristic for a given species, in aspecific geographical area and time frame. Conse-quently, when comparing the tree-ring series (TRS)from a wooden sample of unknown date and originwith master chronologies of the same species repre-senting different geographical areas and covering theperiod when the tree grew, there will be a uniqueposition in which the researched series will match thechronologies of a specific area. Thus, an exactcalendar year can be assigned to each ring of thesample.

bs_bs_banner

The International Journal of Nautical Archaeology (2013) 42.1: 118–136doi: 10.1111/j.1095-9270.2012.00361.x

© 2012 The Authors. International Journal of Nautical Archaeology © 2012 The Nautical Archaeology Society.Published by Blackwell Publishing Ltd. 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.

Precise felling dates help to infer construction datesand allow us to place ship-finds in their historicalcontext. Furthermore, the provenance of the wood canbe deduced from the area represented by the chronolo-gies providing the best statistical matches (dendro-provenancing). This has proved of particularimportance in the field of nautical archaeology, allow-ing us to examine the origins of ships, which by theirvery nature are portable antiquities par excellence.Tree-ring analysis of the hull timbers of the five Skul-delev wrecks provide an exemplar case study on howthe scientific technique can address questions of origin,date, and repair (Bonde and Crumlin-Pedersen, 1990;Crumlin-Pedersen and Olsen, 2002). Recent articleswithin this journal (for example Daly, 2007; Daly andNymoen, 2008) highlight the need for thoughtful inter-pretation of the results of dendrochronological studiesof ships and the potential for error when equatingorigin of timber with origin of ship construction.

For dendrochronology to be successful, however, anetwork of well-replicated ring-width master-chronologies (absolutely anchored in time) must beavailable for the tree species investigated, the region ofprovenance of the wood and the period of interest.While this condition is met in many parts of centraland northern Europe (including most of France,Ireland, Great Britain, the Netherlands, Germany,Scandinavia and the Baltic countries) for certain time-periods, this is not the case for the Iberian Peninsula.In an attempt to redress this imbalance, the project‘Filling in the blanks in European dendrochronology’funded by the Netherlands Organization for ScientificResearch (NWO) was launched in 2009. Part of thisproject aimed to identify ship hull-assemblages in or(expected to be) from Iberia which might benefit fromdendrochronological analysis and/or provide tree ring-width data to assist in the construction of long-spanregional chronologies. This approach follows, to anextent, the successful strategy employed by Guibal andothers in the development of long chronologies for theMediterranean region using in situ Mediterranean shipsites (see Guibal and Pomey, 2003).

The numerous ship-timbers held in store at thefacilities of the Divisão de Arqueologia Náutica e Sub-aquática, Instituto de Gestão do PatrimónioArqui-tectónico e Arqueológico (DANS/IGESPAR) inLisbon, Portugal, provided an opportunity to examinea substantial timber collection from a number ofship sites of note, and to assess their suitability fordendrochronological analysis. In October 2010, M.Domínguez-Delmás and N. Nayling visited the facili-ties of the DANS/IGESPAR and, with the kind assis-tance of their staff, examined, assessed and samplednumerous timbers (Fig. 1). The hull assemblage fromthe Arade 1 wreck proved the most productive wreck-site group, containing numerous substantial ceilingand hull planks, as well as framing elements in goodcondition, with a sufficient number of rings for dendro-chronological analysis and retention of either partial or

complete sapwood (outermost part of stems andbranches, clearly visible in species from the genusQuercus among others) in several timbers. This pro-vided the potential for reasonably constrained felling-date ranges for the parent trees, in some cases accurateto the year or even the season. A second sampling oftimbers from this wreck, directed specifically atframing elements, was conducted in September 2011 byM. Domínguez-Delmás, in order to complete a groupof samples representing all structural parts of theship.

The Arade 1 wreckThe Arade 1 shipwreck was originally discovered in1970 by recreational divers in the Arade River, on thesouthern coast of Portugal (Fig. 2). Reported missingby 1972, the remains were relocated in 2001, by theformer Centro Nacional de Arqueologia Náutica e Sub-aquática (CNANS, now DANS) (Castro, 2006). This

Figure 1. Inspection and selection of timbers for dendro-chronological research at the DANS in Lisbon. (N. Nayling)

Figure 2. Location where the Arade 1 wreck was found atthe mouth of the Arade River, in Portugal. Red star: Lisbon.(V. Loureiro)

M. DOMÍNGUEZ-DELMÁS ET AL.: DENDRO DATING AND PROVENANCING THE ARADE 1

© 2012 The Authors. International Journal of Nautical Archaeology © 2012 The Nautical Archaeology Society 119

was followed by five field missions (Castro, 2002; Riethet al., 2004; Alves et al., 2005; Loureiro and Alves,2005, 2006), including the dismantling and recovery ofthe upper portion of the hull in 2003 (Rieth et al.,2004), and the central section and the lower hull in2005 (Loureiro and Alves, 2005, 2006, 2008). Devoidof significant artefacts, this wooden vessel caught theattention of national and international researchers forits characteristics and its excellent state of preserva-tion. More than five years of post-excavation studyrevealed a skeleton-first, carvel-planked vessel (wherethe frame was used as reference during the shipbuildingprocess, as opposed to the shell-first principle wheremajor importance is given to the hull), planned from aset of pre-established key dimensions and the evolutionof predetermined frames. The latter implies that thetransversal structure was made of two sets of frames,distinct in shape and function: the filling timber frames,Y-shaped, which held and closed the extremities of theship, and the pre-designed frames which defined theshape of the central part of the hull (Loureiro, 2011).The pre-designed frames, on the posterior side (facingthe master frame), have carvings which served as ref-erence points in the construction process, revealingthat frames forward and aft were designed using themaster frame as reference but modified in three ways:raising and narrowing the floor and adjusting the shapeof the futtocks (Loureiro, 2011).

Previous analyses of wood elementsSince the Arade 1 wreck was first found in 1970 a seriesof analyses including radiocarbon and dendrochrono-logical dating, as well as identification of wood speciesof several timber elements, have been carried out ondifferent occasions.

14C and dendrochronological datingBetween 1972 and 2003, radiocarbon (14C) dates of fivedifferent wooden elements have been published, pro-viding a potential construction date anywhere betweenthe second half of the 15th century and the first half ofthe 17th century (Loureiro and Alves, 2008). However,in addition to the imprecision of the 14C results, whichis inherent to the nature of the method (Mitchels,1973), the published 14C reports do not specify where inthe timbers the samples for radiocarbon dating weretaken. This information is crucial for the interpretationof the results, as a sample taken close to the pith of atree that was centuries old when it was cut will providea date range that will be hundreds of years earlier thanthe felling date of the tree. Without a more precisedating of well-selected timber elements, the accuratechronological placement of the Arade 1 wreck with itsassociated shipbuilding characteristics would remainproblematic.

In 2005, T. Wazny, from the Institute for the Study,Conservation and Restoration of Cultural Heritage atNicolaus Copernicus University in Torun, Poland, had

the chance to examine and sample timbers from thewreck for dendrochronological research. His researchon five timbers (planks, filler pieces and a patch) iden-tified by him as deciduous oak (Quercus subg. quercus)resulted in the dating of a filler piece (TAB3) and aplank from an unknown element (A1-X005), with thedates of their outermost surviving rings in AD 1576and AD 1546 respectively (Wazny, unpublishedreport). Whereas sapwood was absent in the sampleA1-X005, it was present on the sample TAB3, allowingan estimation of the felling date of the parent treebetween AD 1577 and AD 1589 (Wazny, unpublishedreport). Given that only two of the samples had dated,Wazny pointed out the need to examine and dendro-chronologically research more timbers containingsapwood, in order to determine whether the ones datedby him belong to the original structure or representrepairs or re-use.

Identification of wood speciesAfter the field campaigns of 2002 and 2005, somesamples were sent for determination of wood species tothe Centro de Investigacão em Paleoecologia Humanaof the former Instituto Português de Arqueologia inLisbon. In 2002, the only sample analysed was identi-fied as Quercus faginea (commonly known as Portu-guese oak) by Van Leeuwaarden and Queiroz (2002).In 2005, another 18 elements were investigated. Fifteensamples were identified as Q. faginea, two as Q. suber(cork oak) and one as Quercus subg. quercus (decidu-ous oak) (Queiroz et al., 2005). These are all speciescommonly found in the Iberian Peninsula. Hence thewood identifications, together with the shipbuildingmethod and other construction characteristics of theship, were taken as support for the hypothesis of a localor regional origin for the ship’s construction (Castro,2006; Loureiro and Alves, 2008). Such accurate iden-tification of the oaks as Q. faginea (down to the specieslevel) is controversial though and requires some con-siderations, which are outlined in the following.

The group of deciduous oaks (Quercus subg.quercus) is represented in Iberia by the species Q. robur,Q. petraea, Q. pubescens, Q. faginea, Q. canariensis andQ. pyrenaica (Amaral Franco, 1990) and their numer-ous hybrid taxa. The differentiation of the living treesof these species and some of their hybrids can be doneby the morphology of their leaves and acorns, but theirwood anatomical features are too similar to discernone species from another (Schweingruber, 1990, 1993).In the late 1990s, Feuillat et al. (1997) announced amethod to distinguish samples from Q. robur and Q.petraea. They reported that for trees of a cambial agebetween 60 and 120 years old, rings c.2 mm wide in Q.robur would present more than two rows of earlywoodvessels and a percentage of latewood less than 60% ofthe total ring-width, whereas in Q. petraea such widerings would present less rows of earlywood vessels anda percentage of latewood higher than 70%. However,this observation only proved valid for 80% of the

NAUTICAL ARCHAEOLOGY, 42.1

120 © 2012 The Authors. International Journal of Nautical Archaeology © 2012 The Nautical Archaeology Society

samples. Furthermore, the limitations of this methodwhen working with (pre)historical timbers are obvious,since samples may not contain such wide rings and inmost cases, the pith and the bark are not present in thesamples, therefore the estimation of the cambial age isdifficult (if not impossible). Hence, the differentiationof the wood from these two species still remains achallenge. In the past decade, a new attempt to identifyanatomical characteristics that would allow the sys-tematic differentiation between, among others, decidu-ous oak species present in Portugal was carried out atthe Centro de Investigacão em Paleoecologia Humanafrom the IGESPAR by the group of P. Queiroz (pers.comm. Queiroz). However, before a significant numberof samples from deciduous oaks was analysed, thestudy was abandoned. Therefore, despite some prom-ising preliminary results, the research remained incon-clusive (pers. comm. Queiroz). A last controversialpoint regarding the identification of Q. faginea is thatits wood has been described as ring-porous to semi-ring porous in forest-stands in the centre of Spain(García Esteban et al., 2003) and as ring-porous inPortuguese forests (Carvalho, 1997). This indicates ahigh variability in the anatomy of the species depend-ing on site conditions, further complicating the poten-tial for discriminating between different species ofdeciduous oak on the basis of wood anatomy.

In view of these considerations, and taking intoaccount the implications of such precise wood identi-fication for interpretation of the potential constructionarea of the ship, we decided to re-examine the group ofsamples identified by Queiroz and others in 2005.

The goals of our study were to find out the precisedate and provenance of timbers used in the construc-tion of the Arade 1 through a dendrochronologicalapproach. The absolute dendrochronological dating ofwell-selected timbers from the wreck could anchor theconstruction of the ship in time, providing a precisechronology for her structural characteristics. Discern-ing the provenance of the wood could shed some lighton the potential geographical origin of the ship or thetrade in timber for her construction. Additionally, theidentification of the wood species of a large number oftimbers of different types would increase our knowl-edge of the selection of species related to their techno-logical properties and according to the function ofspecific elements. Furthermore, it could provide infor-mation about availability and supply, repairs andre-use of timbers.

Selection and sampling of ship-timbersTo be considered suitable for dendrochronologicalresearch, the selected timbers must first of all be of awood species that produces clearly distinct annualrings. This is not always the case for evergreen oakspecies found in the Iberian Peninsula such as Q. suber(Sousa et al., 2009) or Q. ilex (Cherubini et al., 2003;Campelo et al., 2007), nor for tropical species used in

shipbuilding after the establishment of Europeancolonies in the Americas, Africa and Asia. The deter-mination of the wood species may also provide the firstclue towards the provenance of the wood (Giachi et al.,2003; Guibal and Pomey, 2003; Wicha et al., 2003),therefore it is a crucial step when researching ship-wrecks. Hence, our first task while inspecting thetimbers was to identify groups of species (tropicalwood, evergreen oaks, deciduous oaks, others), dis-carding at once those from groups presenting limita-tions for dendrochronological dating. The next stepwas evaluating by roughly counting the number ofrings contained in the timbers, as samples for dendro-chronological research must have a sufficient numberof tree-rings to allow statistically sound results (pref-erably more than 80). And finally, to obtain the fellingdates of the parent trees (or to be able to estimatethem), priority was given to timbers presentingsapwood, or even better, bark-edge. In the absence ofsamples with bark or bark-edge, sampling severaltimbers with sapwood may help narrowing the esti-mated felling-date range of the trees to within a fewyears. In such case, exceptionally, some timbers withfew rings (less than 50 for example) but retainingsapwood, can also be considered for dendrochrono-logical research. In addition, it is desirable to sample asmany suitable timbers as possible, preferably from ele-ments that are likely to belong to the original ship-structure, in different parts of the ship (for example thekeel, floor timbers, futtocks and planks). In this way,the chances of establishing relative dates between(presumably) original timbers is increased, and object-chronologies can be constructed. These object-chronologies are easier to date than tree-ring seriesfrom single samples, as they contain a stronger envi-ronmental signal (Fritts, 1976; Hillam, 1979).

Taking into account all these aspects, we examinedthe sides of the timbers in search of sapwood and thetransverse sections at the ends of planks and framingelements, where the tree-rings and the medullary raysare usually visible. We selected a total of 42 timbers (allplanking and framing elements), including six elements(two planks and four frames) previously researched forwood identification by Queiroz et al. (2005). The woodof 40 elements was identified at once as deciduous oak(Quercus subg. quercus), based on the observation ofmacroscopic characteristics of the anatomy in thetransverse section of the timbers, after cleaning thesurface slightly with razorblades and applying chalkpowder. The deciduous oaks are relatively easy to iden-tify with the naked eye, as they show a ring-porousdisposition of earlywood vessels, abrupt transitionfrom earlywood to latewood, clearly visible multiseri-ate medullary rays and clearly distinguishable ringboundaries (Fig. 3a). Among these timbers was a first-futtock (B-17Eb) that had been identified previously byQueiroz et al. (2005) as cork oak (Q. suber). This ever-green oak species has been described as semi-ringporous, presenting a gradual transition between



earlywood and latewood with round, solitary vesselsgradually decreasing in size in the latewood, and largemultiseriate rays (Schweingruber, 1990; Sousa et al.,2009). However, these characteristics were absent in allthe examined timbers.

Two other timbers (first-futtock B-9Eb and second-futtock Ap0Bb) were identified as chestnut (Castaneasativa). The wood anatomy of this species is similar tothat of the deciduous oaks (ring-porous, with clearlydistinguishable ring boundaries, flame-like groups ofpores in the latewood, uniseriate rays) but it differs inthat it lacks multiseriate medullary rays (Schweingru-ber, 1990) (Fig. 3b).

Cross-sections (2–4 cm thick) were manually sawnfrom the timbers selected for dendrochronologicalanalysis. For wood identification, fragments of c.3 cm3

were removed from the timbers that were not selectedfor dendrochronological research and that neededmicroscopic verification. All the samples were stored insealed plastic bags with water to prevent them fromdrying during their transport to the laboratory.

Dendrochronological researchOnce at the laboratory of the Cultural HeritageAgency of the Netherlands, in Amersfoort, the trans-verse faces of the samples were cleaned with razor-blades from the inner to the outermost ring. Chalk wasrubbed into the cleaned surface to enhance the contrastbetween the ring boundaries. When necessary, micro-slides of the transverse, radial and tangential sectionsfrom the samples were prepared for detailed observa-tion of key anatomical features, which were photo-graphed with a Zeiss AxioCam MRc5 connected to aZeiss Stereomicroscope Discovery-V8.

Ring-widths were measured to the nearest 0.01 mmwith a TimeTable measuring device (University of

Vienna, VIAS) coupled with PAST4 software (B.Knibbe, SCIEM). The resulting series were statisticallycompared with each other and with four of the mea-surements obtained by Wazny in 2005 (internal cross-dating). The fifth timber-element studied by Wazny in2005 (A1-X005) could not be confirmed as belongingto the Arade 1 wreck and was therefore excluded fromthis research. Statistical cross-dating was done withPAST4, using Student’s t-values (t) calculated afternormalization of the raw data with Baillie and Pilcher’salgorithm (Baillie and Pilcher, 1973) and the percent-age of parallel variation or Gleichläufigkeit (Gl) (Eck-stein and Bauch, 1969). The combination of these twostatistical tests, supported by a high significance level(P) of the Gl has proven a reliable and effective methodfor dendrochronological dating. Sound statisticalmatches were visually verified comparing the graphsrepresenting the TRS.

The internal cross-dating allows the identification ofgroups of samples derived from trees growing undersimilar environmental conditions (their series willcross-match well, with t > 6, Gl > 63% and P < 0.005,for example) and even samples that originated from thesame tree. Usually, the latter provide very high statis-tical values (for example t > 10 and Gl > 70% with a P< 0.0001), although lower values may also occur, andthe visual inspection of the matching TRS and therespective wooden samples by experienced dendro-chronologists is the key to a final decision. Series fromthe same tree were averaged and a new internal cross-dating was run. Then, similar, cross-matched TRSwere averaged into object-mean curves followinga hierarchical approach as described by Ecksteinet al. (2009). The resulting object-mean curves andthe remaining single TRS series were comparedwith master and local chronologies from western,central and northern Europe available to the authors.

Figure 3. Macroscopic characteristics of wood anatomy from 3a) deciduous oak species (Quercus subg. quercus) (timber-element St6Bb[2] with dendro-code PAR020) and 3b) chestnut (Castanea sativa) (timber-element Ap0Bb with dendro-codePAR130) in the transverse section. Both species present a ring-porous distribution of earlywood vessels and a sharp transitionbetween earlywood (EW) and latewood (LW). However, chestnut lacks the multiseriate rays (indicated by red arrows in 3a)clearly visible by the naked eye in oak species. Chalk has been applied on the surface to facilitate the visualization of anatomicalfeatures. Growth direction towards the right. Green bar: 1 mm. (M. Domínguez-Delmás)



In addition, the tree-ring data were sent to thedendrochronology laboratory at Arkeolan, in Irun(Spain), where J. Susperregi compared it with chro-nologies from Quercus spp. (Q. robur, Q. petraea andQ. faginea) representing the north-east of Spain.Finally, an object-chronology was computed with thematching dated series using the program ARSTAN(Cook and Holmes, 1986). For this, the TRS werestandardized by fitting them to a 53-year smoothingcubic spline (length of the shortest series representingan individual tree), with 50% cut-off variance, whichwe observed acted similarly to a negative exponentialcurve, but seemed more suited for the dynamic growthtrends of the Arade 1 samples (Cook and Peters, 1981).In this manner, low-frequency trends due to age of thetrees, forest-stand dynamics and other natural orhuman disturbances that would reduce the quality ofthe chronology for future dating purposes, werepartially removed from the series, while keeping thehigh-frequency (year-to-year) variation crucial for den-drochronological dating. The standardized (indexed)series were then averaged with a bi-weight robust meaninto the object-chronology.

The provenance of the wood can be inferred fromthe chronologies dating the samples, as long as chro-nologies representing specific areas are used. Ideally,the area represented by the chronologies providingthe best statistical matches will be the area of prov-enance of the wood. But if the wood originates froman area where reference chronologies for the speciesstudied are lacking, the samples will likely remainundated.

ResultsOur final set of samples comprised a total of 58 ele-ments, including the 42 timbers inspected and sampledfor this research (24 for tree-ring research and 18 forwood identification, six of which had already beenidentified by Queiroz and others in 2005), the rest ofthe samples analysed by Queiroz et al. (2005) and fouroak TRS from the timbers previously researched byWazny (two filler pieces, a ceiling plank and a patchfrom a framing element) (Table 1 and Fig. 4).

Wood identificationsThe identifications of 42 selected timber-elementsinspected at the DANS (40 deciduous oaks and twochestnuts), were verified and confirmed at the labora-tory by the observation of wood anatomical featuresunder the microscope. In this way, the identificationsfrom 2005 of the first-futtock B-17Eb and a treenailfrom the ceiling plank A1-151 as cork oak (Q. suber)were definitely discarded (Fig. 5a, b, c). Furthermore,this treenail was a peculiar case, as it had beenwedged tight with a smaller wooden nail driven intoits centre (Fig. 5c). As a result, the nail (made ofdeciduous oak) had caused the vessels, rays and fibresof the treenail to compress around the point of entry,

therefore the wood appeared to have a higher densityof rays (which is a key characteristic of cork oak) inmacroscopic observation. However, the microscopicobservation of the transverse section (Fig. 5b, c)allowed the observation of the multiseriate rays andthe compressed fibres, and made obvious the abruptchange from earlywood to latewood. These featuresled to the identification of the treenail as deciduousoak rather than cork oak.

The observation of key anatomical features on therest of the samples previously researched by Queirozet al. (2005) resulted in the identification of all of themas deciduous oak (Quercus subg. Quercus) (Table 1). Ingeneral, these samples showed differences in growth-rate (some derived from fast-grown trees and othersfrom slow-grown trees) and some presented abundantlarge multiseriate rays (Fig. 6).

Description of samplesFrom the 24 timbers selected for dendrochronologicalresearch, five hull-planks (St5Bb[1], St6Bb[2],St7Bb[1], St6Eb[1] and St5Eb[2]) and five framing ele-ments (A-141, B2Eb, Ap6Bb, A1-237 and A1-10) con-tained partial sapwood, whereas two first-futtocks(B-8Eb, B-16Eb) and two second-futtocks (Ap5Bb,A1-239) retained complete sapwood and bark-edge(Table 1). The tree used to make the second-futtockA1-239 was felled in late summer or winter (Septemberto March of the following year) as inferred from thecompleteness of the outermost ring, whereas the otherthree trees were felled during the growing season(somewhere between April and June), as implied by thepresence of an incomplete outermost ring at the bark-edge. Furthermore, these samples presented numeroussapwood vessels filled with tyloses (Fig. 7), a commonreaction of oak trees that have been cut duringthe growing season (Murmanis, 1975; pers. comm.F. Schweingruber).

All the planks had been tangentially converted,suggesting they were sawn (as opposed to split orcleaved) from the log. The framing elements were allhalved or quartered timbers. Their sizes, togetherwith the presence of sapwood and pith, allowedan estimation of the diameter of the parent trees of200–300 mm.

Tree-ring typology: fast v. slow growth-ratesThe TRS from the researched timbers were of differentlengths and, although most of them presented similargrowth patterns, with slow growth-rates after theinitial fast-grown juvenile period (30–40 years),the planks presented more regular patterns than theframing-elements, which showed dynamic ones. Someplanks, however (hull-planks St6Eb[1], St5Eb[2] andSt2Eb[3], filler pieces TAB3 and TAB2, and ceilingplank TN4) showed fast growth-rates, with mean ring-width values close to or more than 1.50 mm per yearand two of them having mean year growths more than2.20 mm (Table 1). The fast-grown trees from which



Tab

le1.

Lis

tof

rese

arch

edel

emen

ts(i

nclu

ding

thos

ere

sear

ched

in20

05),

resu

lts,

woo

did

enti

ficat

ion

and

desc

ript

ion

ofsa

mpl

es;

A:

rese

arch

edby

Que

iroz

etal

.(20

05)

for

woo

did

enti

ficat

ion

(ind

icat

ing

inbr

acke

tsth

esa

mpl

enu

mbe

r);

B:

rese

arch

edby

Waz

nyin

2005

for

woo

did

enti

ficat

ion

and

dend

roch

rono

logi

cal

dati

ng;

Q:

Que

rcus

subg

.qu

ercu

s;C

:C

asta

nea

sati

va;

N:

num

ber

ofri

ngs

inth

esa

mpl

e;P

ith:

pres

ent

(+)

orab

sent

(-);

SW

:sa

pwoo

d;M

RW

tota

l:m

ean

ring

-wid

th(m

m)

tota

lle

ngth

seri

es;

MR

W(-

35):

mea

nri

ng-w

idth

(mm

)ex

clud

ing

35ye

ars

juve

nile

woo

d.S

hade

dro

ws:

sam

ples

wit

hba

rk-e

dge

pres

ent.

*The

exac

tty

peof

elem

ents

coul

dno

tbe

iden

tifie

d

Sam

ples

from

prev

ious

stud

ies

Den

dro-

code

Ele

men

tco

deT

ype

Fun

ctio

nSp

ecie

sN

Pit

hSW

MR

Wto

tal(

mm

)M

RW

(-35

)(m

m)

—P

AR

010

St1B

b[2]

Hul

lpla

nkH

ull

Q13

1+

—1.

260.

93—

PA

R10

0St

8/9B

bH

ullp

lank

Hul

lQ

103

-—

1.12

1.00

—P

AR

030

St5B

b[1]

Hul

lpla

nkH

ull

Q12

1-

51.

150.

98—

PA

R09

0A

1-24

2P

lank

*U

ndet

erm

ined

Q10

1+

—1.

441.

21—

PA

R02

0St

6Bb[

2]H

ullp

lank

Hul

lQ

130

+11

1.24

1.09

—P

AR

040

A1-

258

Pla

nkC

eilin

gQ

74-

—0.

820.

82—

PA

R05

0A

1-61

Hul

lpla

nkH

ull

Q78

-—

1.18

1.18

—P

AR

060

St7E

b[2]

Hul

lpla

nkH

ull

Q13

8-

—0.

870.

87—

PA

R07

0St

7Bb[

1]H

ullp

lank

Hul

lQ

152

+5

1.01

0.67

A(9

)P

AR

230

St6E

b[1]

Hul

lpla

nkH

ull

Q56

-(+

4s.b

.)2.

232.

23—

PA

R24

0St

11B

bH

ullp

lank

Hul

lQ

102

+10

—1.

411.

17B

PT

0010

5T

AB

3F

iller

piec

eC

eilin

gQ

49-

141.

801.

80B

PT

0010

4T

AB

2F

iller

piec

eC

eilin

gQ

41-

161.

471.

47—

PA

R08

0St

5Eb[

2]H

ullp

lank

Hul

lQ

47-

62.

632.

63—

PA

R22

1St

2Eb[

3]H

ullp

lank

Hul

lQ

83-

—1.

721.

72B

PT

0010

2T

N4

Cei

ling

plan

kC

eilin

gQ

79-

—1.

711.

71B

PT

0010

3C

-10

Pat

chof

C11

floor

tim

ber

Fra

me

Q11

2-

—0.

790.

79

—P

AR

110

B-8

Eb

Fir

st-f

utto

ckF

ram

eQ

126

-21

(+1)

0.95

0.80

—P

AR

120

B-9

Eb

Fir

st-f

utto

ckF

ram

eC

92-

—1.

581.

58—

PA

R13

0A

p0B

bSe

cond

-fut

tock

Fra

me

C15

2-

—1.

011.

01—

PA

R14

0A

1-14

1Se

cond

-fut

tock

Fra

me

Q91

+25

1.17

0.77

—P

AR

150

Ap5

Bb

Seco

nd-f

utto

ckF

ram

eQ

142

+60

(+1)

0.77

0.58

—P

AR

160

B2E

bF

irst

-fut

tock

Fra

me

Q88

-23

1.60

1.60

—P

AR

170

Ap6

Bb

Seco

nd-f

utto

ckF

ram

eQ

138

+14

0.92

0.91

—P

AR

180

A1-

237

Fra

me*

Fra

me

Q13

9+

8(+6

)0.

900.

84—

PA

R19

0A

1-23

9Se

cond

-fut

tock

Fra

me

Q70

+24

1.27

1.23

—P

AR

200

A1-

10U

ndet

erm

ined

Fra

me

Q73

+521

1.15

0.85

A(5

)P

AR

210

B-1

6Eb

Fir

st-f

utto

ckF

ram

eQ

106

+45

(+1)

1.06

0.69



Tab

le1.

Con

tinu

ed

Sam

ples

from

prev

ious

stud

ies

Den

dro-

code

Ele

men

tco

deT

ype

Fun

ctio

nSp

ecie

sN

Pit

hSW

MR

Wto

tal(

mm

)M

RW

(-35

)(m

m)

——

St1E

bH

ullp

lank

Hul

lQ

—-

——

—A

(6)

—A

1-14

1T

reen

ail

Fra

me

Q—

-—

——

A(8

)—

A1-

151

Cei

ling

plan

kC

eilin

gQ

—-

——

—A

(1)

—A

1-15

1T

reen

ail

Cei

ling

Q—

-—

——

——

C5

Flo

or-t

imbe

rF

ram

eQ

—-

——

——

—B

1Eb

Fir

st-f

utto

ckF

ram

eQ

—-

——

——

—B

4Bb

Fir

st-f

utto

ckF

ram

eQ

—-

——

——

—A

1-18

5Se

cond

-fut

tock

Fra

me

Q—

-—

——

——

A1-

195

Seco

nd-f

utto

ckF

ram

eQ

—-

——

——

—C

9F

loor

-tim

ber

Fra

me

Q—

-—

——

——

A1-

248

Seco

nd-f

utto

ckF

ram

eQ

—-

——

—A

(14)

—C

-19

Flo

orti

mbe

rF

ram

eQ

—-

——

——

—C

-18

Flo

orti

mbe

rF

ram

eQ

—-

——

—A

(3)

—B

-17E

bF

irst

-fut

tock

Fra

me

Q—

-—

——

——

B-1

8Eb

Fir

st-f

utto

ckF

ram

eQ

—-

——

—A

(2)

—A

1-64

Hul

lpla

nkH

ull

Q—

-—

——

A(4

)—

RP

rSt

empo

stL

ongi

tudi

nal

Q—

-—

——

A(7

)—

B-6

Eb

Fir

st-f

utto

ckF

ram

eQ

—-

——

—A

(10)

—A

1-15

8St

ring

erC

eilin

gQ

—-

——

—A

(13)

—St

7Eb(

1)H

ullp

lank

Hul

lQ

—-

——

—A

(11)

—St

7Eb(

1)T

reen

ail

Hul

lQ

—-

——

—A

(12)

—T

n3A

/BC

eilin

gpl

ank

Cei

ling

Q—

-—

——

A(1

5)—

Tab

1F

iller

piec

eC

eilin

gQ

—-

——

—A

(16)

—T

ab4

Fill

erpi

ece

Cei

ling

Q—

-—

——

A(1

7)—

C-1

9T

reen

ail

Flo

or-t

imbe

rQ

—-

——

—A

(18)

—B

-16E

bT

reen

ail

Fir

st-f

utto

ckQ

—-

——

——

—B

-14E

bF

irst

-fut

tock

Fra

me

Q—

-—

——

——

C-5

Flo

orti

mbe

rF

ram

eQ

—-

——

——

—B

3Bb

Fir

st-f

utto

ckF

ram

eQ

—-

——

——

—B

4Eb

Fir

st-f

utto

ckF

ram

eQ

—-

——

—



Figure 4. Researched timber-elements in the Arade 1 wreck. (V. Loureiro and J. Gachet Alves)



these planks were sawn, grew in favourable conditions,without shortage of water or lack of light. Three ofthese samples (hull-plank St5Eb[2] and the filler piecesTAB3 and TAB2 researched by Wazny in 2005) pre-sented less than 50 rings, but had also been selected forthe research because they contained sapwood. Exclud-ing the fast-grown planks and the first 35 rings of juve-nile wood in the rest of the planks containing the pith,the mean ring-width was found to drop down to0.98 mm per year (Table 1). Such slow growth is char-acteristic of oak trees living under limiting conditions(for example, competition for light when growing in adense forest).

Most of the framing elements contained the pith andshowed similar growing conditions to that of the slow-grown trees used for planking (juvenile period ofdynamic growth, followed by a growth reduction).Applying the same criterion of exclusion of juvenilewood as with the planks, the mean ring-width value offraming elements is 0.97 mm per year. Having estimated

the diameters of these elements containing pith andsapwood between 200–300 mm and considering theirnumber of tree-rings (between 70 and 142), we can alsodescribe them as derived from slow-grown trees.

Dating results and provenanceThe dendrochronological research included four TRSobtained by Wazny in 2005 and 24 series obtainedfrom the new samples. The first internal cross-dating(comparison of all TRS among them) resulted in theidentification of planking elements derived from thesame tree. The outstanding statistical results betweenthe pairs of TRS from hull-planks St1Bb[2] and St8/9Bb (dendro-codes PAR010-PAR100, t = 8.44 andGl = 75.7 with P < 0.0001), St5Bb[1] and A1-242(PAR030-PAR090, t = 10.7, Gl = 85.4% and P <0.0001), hull plank St2Eb[3] and ceiling plank TN4(PAR221-PT00102, t = 11.2 and Gl = 83.3 with P <0.0001) confirmed by visual match, indicate that the sixplanks were sawn from three trees.

Figure 5. Transverse section of wood from first-futtock B-17Eb (5a) and a treenail from the ceiling plank A1-151 (5b) withwooden nail inserted in the middle (5c). Both samples, as well as the nail inserted into the treenail, present anatomicalcharacteristics typical for deciduous oaks (Quercus subg. quercus) described in the text. Red arrows point at multiseriate rays.Green bar: 1 mm. (M. Domínguez-Delmás)



Furthermore, the plank St5Eb[2] (PAR080) gave anoutstanding visual match with the filler pieces TAB2(PT00104A) and TAB3 (PT00105A) researched byWazny in 2005, indicating also that the three timbersderived from the same tree. This observation,although not sustained by outstanding statisticalvalues as a result of the short overlaps (mean groupoverlap = 42.3 rings, t = 5.25 and Gl = 76.2), is sup-ported by the fact that all three planks are sawn fromthe same tangential axis of the stem, and thereforehave a similar shape, growth pattern and number ofrings. These TRS of samples from the same tree wereaveraged into the tree mean-curves PARQT1(PAR010-PAR100), PARQT2 (PAR030-PAR090),PARQT3 (PAR080-PT00104A-PT00105A) andPARQT4 (PAR221-PT00102).

The second internal cross-dating, including thesetree-mean curves and the rest of the TRS, resulted in

the relative dating of almost the entire group of timbers(Table 2). The results on Table 2 show high statisticalvalues between most of the planking elements amongthem (PARQT1, PARQT2, PAR020, PAR040,PAR050, PAR060, PAR070, PAR230 and PAR240)and with some of the framing elements (PAR110,PAR140, PAR150 and PAR190) which also showedexcellent correlation values between them. This indi-cates that these oak timbers represent a rather homo-geneous group. Therefore it was decided to join allthese TRS into a tentative object-mean curve(PARQMC), which was then compared with theremaining TRS. PT00103A and PARQT3 were finallyadded to the object-mean curve, as they still correlatedwith highly significant statistical values (t = 4.18, Gl =65.2, P < 0.001 and t = 4.01, Gl = 71.7, P < 0.001respectively). The statistical values of the initial relativematch of PARQT4 with PAR050 (see Table 2)decreased drastically when compared to the object-mean curve (t = 1.95, Gl = 62.7, P < 0.005), hence it wasexcluded from the mean. Likewise, the TRS from oakframing elements PAR170, PAR200 and PAR210 wereleft out of PARQMC.

The TRS from the chestnut timbers showed a goodsynchronization between them (t = 4.74, Gl = 66.3, P <0.005) and with some TRS from oak elements(Table 2). The chestnut series were averaged into themean curve PARCSM.

The comparison of the oak and chestnut meancurves (PARQMC and PARCSM respectively) andthe remaining unassigned TRS (PARQT4, PAR170,PAR200, PR210) with the master and local chronolo-gies resulted in the dating of all the TRS except forPARQT4 (Fig. 8). Framing elements containing bark-edge provided precise felling dates: PAR210, spring/summer AD 1579; PAR190, late summer/winter AD1582/83; and PAR110 and PAR150, spring/summerAD 1583. To estimate the felling dates of sampleswith partial sapwood or the terminus post quem forsamples without sapwood, sapwood estimations fromthe source area should be used. Regional chronologiesdating the series represented predominantly the westof France. Given that sapwood statistics for this areaare lacking, we used sapwood estimates proposed byPilcher (1987) for northern France. Pilcher calculatedthat the number of sapwood rings in oaks growing inthe north of France can be expected to range between15.25 and 43.26 within a 95% confidence interval.Applying this sapwood estimate, the calculatedfelling-date ranges for the samples with partialsapwood were found to span dates between AD 1551and AD 1608 (Fig. 8). For the samples lacking anysapwood only terminus post quem dates can be pro-vided (dates after which the trees were felled).However, considering the homogeneity of theresearched material and most of the felling-dateranges of the samples with partial sapwood, it seemsvery likely that all timbers were felled between the late1570s and AD 1583.

Figure 6. 6a) Transverse section of sample from timber-element B-8Eb (dendro-code PAR110), showing large andabundant multiseriate rays; 6b) Sample A1-237 (dendro-codePAR180) used for comparison. Chalk has been applied onthe surface to facilitate the visualization of anatomical fea-tures. Green bar: 1 mm. (M. Domínguez-Delmás)



Figure 7. Cross-sections of samples from framing-elements A1-141 (7a, dendro-code PAR140), Ap5Bb (7b, PAR150) andB-16Eb (7c, PAR210) with detail of a portion of the sapwood with vessels free of tyloses (7a) and with numerous vesselsabnormally filled with tyloses (7b,c). Black and white scale represents 200 mm, red bar 10 mm and green bar 1 mm. (M.Domínguez-Delmás)



Tab

le2.

Res

ults

seco

ndin

tern

alcr

oss-

dati

ng.D

endr

o-co

des

repr

esen

tin

divi

dual

tree

s.#

:P

<0.

05;

##

:P

<0.

01;

##

#:

P<

0.00

1;#

##

#:

P<

0.00

01;

n.s.

:no

tsi

gnifi

cant

;n.

o.:

noov

erla

p.S

tude

nt’s

t-va

lues

low

erth

an3

are

not

show

n



The object-mean curve PARQMC shows excellentmatches (t higher than 9) with local oak chronologiesfrom the Pays-de-la-Loire region derived from timbersof buildings in the area between Nantes, Angers andthe Royal Abbey of Fontevraud (Fig. 9). This indi-cates that the researched timbers very likely originatefrom forest areas along the Loire River, in westernFrance. Furthermore, it also presents high correlationvalues with chronologies further north, indicating thatthe object-mean contains a strong regional signal. Theresulting oak chronology (PARQCR) has beenplotted against the French local chronology (49)-Fontevraud which provided the best statistical match(Fig. 10).

Regarding the TRS not included in the mean,PAR200 showed highly significant correlations withlocal chronologies by or closer to the western coast(Olonne-sur-Mer (85), t = 6.23, Gl = 71.2, P < 0.0005and (85)-Vendée, t = 6.04, Gl = 72.6, P < 0.0002).Similarly, PAR210 and PAR170 showed a good corre-lation with the coastal chronology from Olonne-sur-Mer (t = 4.01, Gl = 64.6, P < 0.01 and t = 4.15, Gl =67.3, P < 0.0005 respectively).

The chestnut mean curve PARCSM also showedexcellent matches (t higher than 5.5) with local oakchronologies from the Pays de la Loire (Fig. 11). Theseheteroconnections served to absolutely date the chest-nut framing elements to the calendar year, althoughthe lack of bark-edge in the samples and visiblesapwood in this species reduced the possibility of ascer-taining the felling date of the parent trees.

DiscussionThe researched samples constitute a homogeneousgroup of timbers, originating from the west of France,very likely from the area around Fontevraud, in thePays de la Loire. This implies that the dated oak timbersbelong to some of the deciduous species predominant inthat area (Q. robur, Q. petraea, or Q. pubescens). Thisregion is outside the natural distribution range of Q.faginea (Tutin et al., 2001), therefore this species can beexcluded as species for the researched timbers. Themacroscopic wood anatomy within a tree species maydiffer not only between sites, depending on environmen-tal conditions such as local disturbances, but also within

Figure 8. Time-span of the dated oaks (yellow) and chestnuts (orange) TRS. The dot indicates that the pith is present in thesample and the grey filling shows the sapwood rings in the oak samples. Clustered bars represent timbers made from the sametree (bar diagram obtained with software Dendro for Windows by Tyers 1997). (M. Domínguez Delmás)



a tree, from juvenile to mature wood or from stem tobranch wood. Differences in growth-rate amongdeciduous oak species are exclusively related to the sitewhere the trees grow and are by no means associatedwith the genetics of certain species. The presence ofabundant large rays is indeed a key characteristic ofevergreen oak species such as Q. ilex or Q. suber, but thisis also an anatomical peculiarity frequent in the prox-imities of branches and other protuberances in the woodof other oak species (pers. comm. F. Schweingruber),hence it could indicate that the timber was obtainedfrom a branch, instead of stem-wood. However, as ofyet, no comprehensive studies are available on thissubject and no conclusions can be drawn in this respect.Nevertheless, considering that certain timbers for ship-building were specifically selected from branches orfrom the part of the stem where the branches fork(Albion, 1926; De Aranda y Antón, 1990, 1999), itshould not be surprising to find such aberrant rays.Therefore caution should be the norm when using themas a key to identify the wood down to the species level inthe case of deciduous oaks.

High similarity between the chestnut and oak TRS(heteroconnection) indicates that the trees of bothspecies selected for ship-timber must have beengrowing under similar conditions, very likely in thesame area. This is very plausible, as the distribution of

some deciduous oak species overlaps broadly with thatof chestnut in the west of France (Fig. 12). Heterocon-nections between oak species and chestnut from thesame region have been observed before in the Pays dela Loire in construction timbers from the Château desDucs de Bretagne at Nantes and the cloister of theconvent of Château-Gontier for example, but also onpainting panels made of chestnut at Limoges, whichcould be dated with local oak chronologies (C. Lavier.pers. comm.). These examples reinforce the idea thatthe chestnut and oak trees from this study weregrowing in close proximity.

The tree-ring patterns of the timbers presentingpith and sapwood (fast grown in the juvenile periodwith a decrease in growth-rate shortly afterwards), areprobably the result of the trees growing under a rela-tively closed canopy (in a relatively dense forest). Suchgradual decrease in growth does not seem to be due tothe natural age-trend of trees, given that these treeswere still relatively young (70 to 142 years), thereforethey might be due to light competition with adjacenttrees (canopy closure) as the trees matured. The fast-grown pattern found in two planks, is common fortrees growing from coppice-stools or in relatively openlandscapes or coastal areas, for example, where theavailability of light or water is not a limiting factor fortree growth (Schweingruber, 1996; Beeckman, 2005;

Figure 9. Map showing the spatial distribution of t-valuesresulting from cross-dating the oak object-chronologyPARQMC with local oak chronologies from France andBelgium. (C. Lavier)

Figure 10. Visual synchronization between the standardized oak object-chronology PARQCR (yellow) and the French localchronology (49)-Fontevraud (black); X-axis: calendar years; Y-axis: dimensionless ring-width indices. (M. Domínguez-Delmás)

Figure 11. Spatial distribution of t-values resulting fromcross-dating the chestnut mean curve PARCSM with localoak chronologies from France. (C. Lavier)



Haneca et al., 2006). The fact that the TRS from thefast-grown trees match the ones from the slow-grownones and that there is not a division of timber-elements based on growth-rates, but rather on treesize, indicates that the selection of trees was restrictedto a certain geographical area with variable coverdensity (some parts would be more open than others).In the past decade, combined dendrochronologicaland historical research in eastern France (Beck et al.,2002) demonstrated that in Burgundy, for example,from the middle ages until the beginning of the

Modern Period, construction timber for buildings wassupplied from areas within a distance of 10 to 50 km.The selection of such construction timber was basedon quality (sound wood free of rot, wounds, fungi),shape (straight, forked, curvy) and diameter depend-ing on the needs, but the wood usually originatedfrom the same forest areas (pers. comm. C. Lavier;Locatelli et al., 2011). Similarly, the homogeneityin the sizes of the framing elements points at a selec-tion of trees based on diameter and shape. The planksoriginated from bigger trees, as inferred from theestimated minimum size of the few ones containingpith and sapwood (400–500 mm diameter). Findingwood derived from the same tree in different typesof elements (as in the case of the filler piecesTAB2, TAB 3 and the hull-plank St5Eb[2], or the hullplank St2Eb[3] and ceiling plank TN4) suggeststhat wood was transported to the dockyard as fulllogs, to be processed there into the required shapes.

Although the number of selected timbers for den-drochronological research represents a low percentageof the total number of recovered wooden elements,their different timber types and wide distributionwithin the ship makes them representative. The homo-geneity in provenance of the wood, along with precisefelling dates (and estimated ones) for the oak timbersranging between AD 1579 and 1583, indicates that theymost likely belong to the original construction of theship. Different felling dates point to stock-piling prac-tices, as the timbers did not show signs of re-use. Theconstruction of the ship could then be placed shortlyafter AD 1583, as seasoning time for the wood beforeits use as ship-timber has been described as short(Dodds and Moore, 1984). An alternative, but lesslikely interpretation of the results, is that the ship wasconstructed soon after AD 1579 and that the timbersfrom trees felled around AD 1583 represent repairs.The fact that chestnut wood was used in framing ele-ments could indicate either that those futtocks wererepaired or that oak and chestnut were used togetherbecause they were available. Guibal and Pomey (2003)argue that the homogeneity of timbers is a reflection ofquality and an effective and abundant supply of theraw material. In this study, the lack of precise fellingdates for the chestnut trees makes it impossible toascribe them to a specific event (original construction,re-use or repair).

The place where the ship was built remains unde-termined. The high strength and durability of oakwood made it a preferred species for structural shipelements (Albion, 1926; De Aranda y Antón, 1999).Its limited availability in the more populated and lessforested areas of western Europe prompted its trade(for shipbuilding and other uses) during the earlymodern period from forested areas in north-easternEurope, mainly Germany, Scandinavia, Poland andother Baltic countries (Wazny and Eckstein, 1987; DeVries and Van der Woude, 1995; Delmás and Van denBerselaar, 2009). This trade was for a large part

Figure 12. Distribution maps of Quercus robur (a), Q.petraea (b) and Castanea sativa (c) (source: EUFORGEN2009, www.euforgen.org)



carried out by Dutch and Flemish merchants, as wellas Hanseatic traders, who served as a link betweennorth-eastern and south-western Europe (De Vriesand Van der Woude, 1995; Delmás and Van den Ber-selaar, 2009). In Iberia, documentary evidence for thistrade can be found in Cantabrian and Basque archivesfor the 15th and 16th centuries (Casado Soto, 1998;Orella Unzué, 2003) and for the 16th century in thenorth of Portugal, Lisbon and Seville, for example,where oak timber was imported from the Baltic to beused in shipbuilding and art-pieces (Devy-Vareta,1986; Bruquetas, 2002; Klein and Esteves, 2001; OtteSander, 2008). Furthermore, Devy-Vareta (1986)describes the import of French timber to harbours inthe north-west of Portugal from the 13th to the 16thcenturies. In the context of such active timber trade,the Iberian construction of the Arade 1 ship cannot beruled out as a working hypothesis. However, it wouldhave to be supported by accurate documentary evi-dence, as the 16th century was a time of diplomaticdevelopments and political instability (especially thelast quarter of the century), hence trade connectionsprevailing in the first decades of the century may havebeen disrupted by the 1580s.

The construction characteristics of the Arade 1alone cannot provide conclusive evidence for the ship’sconstruction location. By the late-16th century, carvelship construction had become widespread on theAtlantic seaboard and in the southern North Sea, withthe shipyard of Rouen having played an important rolein the transfer of Mediterranean nautical technologies(Rieth, 1989). The development and spread of carvel-built vessels is one of the most important questions innautical archaeology and the role of Iberian ships inthis technological expansion has been the subject ofnumerous site-specific explorations, doctoral thesesand symposia (for example Alves, 2001; Castro andCuster, 2008). Many of these studies have focused ondetails of construction, often in an attempt to define aspecifically Iberian-Atlantic shipbuilding tradition.More nuanced approaches, integrating studies of woodscience and documentary sources have advocated, orsought to achieve, synthesis of technological detailwith the evidence for forestry management and timbersupply (for example Grenier et al., 2007; Creasman,2008). Only through such multi-disciplinary

approaches, including research on the dynamics of thetimber trade, can the full complexity of developmentsin shipbuilding traditions at that time in history beunderstood.

ConclusionsDendrochronological research has provided a precisedate and accurate provenance for the researchedtimbers. However, the question of where the ship wasbuilt remains open. The identification of deciduousoak of the type Q. robur/petraea/pubescens and chest-nut originating from the Pays de la Loire, in westernFrance, implies either that the ship was built in aFrench dockyard, or in some other shipyard tradingFrench timber. Further historical research on thetimber trade from France in the late-16th centurycould shed some light on this question.

This study illustrates the potential of dendrochrono-logical research when applied to shipwreck assem-blages through employment of an appropriatesampling strategy, which includes examination of abroad range of structural elements and sampling ofnumerous, well-selected timbers. The existence of adense network of well-replicated local chronologiesavailable for the source area has been the key to thesuccessful dating and provenancing of the timbers. Ifthe timber had originated from areas where such tree-ring networks are lacking (for example the Iberian Pen-insula), the tree-ring series would probably haveremained undated. The development of reference tree-ring chronologies in areas which acted as ship-timbersuppliers is still a much-needed step towards the assess-ment of wooden ship remains all over the world.Timber from shipwrecks is an excellent key for under-standing past shipbuilding traditions and forest man-agement practices, as well as a resource for developingtree-ring width data to improve existing chronologiesand develop new ones. This paper seeks to encouragenautical archaeologists (especially those studying shipremains in areas where these approaches have not beenextensively used) to implement and fully exploit thepotential of wood science through appropriate sam-pling strategies, and contribute in this way to theunderstanding and preservation of this particular formof cultural heritage.

AcknowledgementsThanks to Francisco Alves (ex-director DANS) for welcoming this research and providing the required logistics; João Coelho,Miguel Adolfo Martins (DANS) and Pedro Neves de Oliveira for their invaluable assistance during the inspection and samplingof timbers; Prof. Filipe Castro (Texas A&M University) for setting up the contacts and for his encouragement and enthusiasm;Josué Susperregi (Arkeolan) for checking our data with Basque chronologies; Fritz H. Schweingruber and Paula Queiroz fortheir comments about wood anatomical aspects; and an anonymous reviewer for the helpful comments.

Research carried out within the project ‘Filling in the blanks in European dendrochronology’ funded by the NetherlandsOrganization for Scientific Research (NWO). The Malcolm H. Wiener Foundation provided financial support for TomaszWazny. The compiled data is stored at the DCCD facility (http://dendro.dans.knaw.nl).



ReferencesAlbion, R. G., 1926, Forests and Sea Power: the Timber Problem of the Royal Navy, 1652–1862. Harvard University Press,

Cambridge.Alves, F. (ed), 2001, Proceedings of the International Symposium ‘Archaeology of Medieval and Modern Ships of Iberian-Atlantic

Tradition’. Lisbon.Alves, F., Machado, A., and Castro, F., 2005, Resultados preliminares da campanha de trabalhos arqueológicos Arade 2001,

realizada no âmbito do projecto Pró-Arade, in M. J. Gonçalves (ed.), Actas do 2o Encontro de Arqueologia do Algarve, XELB5, 257–66.

Amaral Franco, J., 1990, Quercus. In G. López, (ed.) Flora Ibérica, vol. II. Real Jardín Botánico de Madrid, Servicio dePublicaciones del CSIC, 15–36, Madrid.

Baillie, M. G. L. and Pilcher, J. R., 1973. A simple crossdating program for tree-ring research. Tree-Ring Bulletin 33,7–14.

Beck, P., Auloy, G., Blondeaux, L., Canat, C., Desvignes, J., Locatelli, C., Maerten, M., Morlon, J. C., Pinette, M., andPousset, D., 2002, Vie de cour en Bourgogne à la fin du Moyen Age. Marguerite de Flandre et Germolles (1380–1405), SuttonEd.

Beeckman, H., 2005, The impact of forest management on wood quality. The case of medieval oak, in C. Van de Velde, H.Beeckman, J. Van Acker, and F. Verhaeghe (eds), Constructing wooden images: Proceedings of the symposium on theorganization of labour and working practices of late gothic carved altarpieces in the Low Countries, 93–113, Brussels.

Bonde, N. and Crumlin-Pedersen, O., 1990, The dating of wreck 2, the long ship, from Skuldelev, Denmark, NewsWARP 7,3–6.

Bruquetas, R., 2002, Técnicas y materiales de la pintura española en los Siglos de Oro, Fundación de Apoyo a la Historia del ArteHispano, Madrid.

Campelo, F., Gutierrez, E., Ribas, M., Nabais, C., and Freitas, H., 2007, Relationships between climate and double rings inQuercus ilex from northeast Spain. Canadian Journal of Forest Research 37.10, 1915–23.

Carvalho, A., 1997, Madeiras Portuguesas. Estrutura anatómica, Propriedades e utilizações. Vol. II, Direcção-Geral das Flo-restas, Lisboa.

Casado Soto, J. L., 1998. Aproximación a la tipología naval cantábrica en la primera mitad del siglo XVI, Itsas Memoria.Revista de Estudios Marítimos del País Vaco, 2, Untzi Museoa-Museo Naval, Donostia-San Sebastián, 169–91.

Castro, F., 2002, The Arade 1 Ship – 2002 Field Season. Vol. 1: The Site – ShipLab Report 3. On file in IPA/CNANS’ library,2002, and in Nautical Archaeological Program Library, Texas A&M University.

Castro, F., 2006, The Arade 1 shipwreck. A small ship at the mouth of the Arade River, Portugal, in L. Blue, F. Hocker andA. Englert (eds), Connected by the Sea. Proceedings of the 10th Symposium on Boat and Ship Archaeology 2003 (ISBSA 10),300–5, Roskilde.

Castro, F. and Custer, K. (eds.), 2008, Edge of Empire: Proceedings of the Symposium ‘Edge of Empire’ held at the 2006 AnnualMeeting of the Society for Historical Archaeology, Sacramento, Sintra.

Cherubini, P., Gartner, B. L., Tognetti, R., Bräker, O. U., Schoch, B., and Innes, J. L., 2003, Identification, measurement andinterpretation of tree rings in woody species from Mediterranean climates. Biology Reviews 78, 119–48.

Cook, E. R. and Holmes, R. L., 1986, Guide for computer program ARSTAN, in R. L. Holmes, R. K. Adams and H. C. Fritts(eds), Tree-Ring Chronologies of Western North America: California, eastern Oregon and northern Great Basin, Laboratory ofTree Ring Research, The University of Arizona, 50–65, Tucson.

Cook, E. R. and Peters, K., 1981, The smoothing spline: a new approach to standardizing forest interior tree-ring width seriesfor dendroclimatic studies, Tree-Ring Bulletin 41, 45–53.

Creasman, P. P., 2008, Ship timber: forests and ships in the Iberian peninsula during the age of discovery, in F. Castro and K.Custer (eds), Edge of Empire: Proceedings of the Symposium ‘Edge of Empire’ held at the 2006 Annual Meeting of the Societyfor Historical Archaeology, Sacramento, 235–48, Sintra.

Crumlin-Pedersen, O. and Olsen, O., 2002, The Skuldelev Ships I: Topography, Archaeology, History, Conservation andDisplay, Ship and Boats of the North 4.1, 360.

Daly, A., 2007, The Karschau Ship, Schleswig-Holstein: Dendrochronological Results and Timber Provenance, IJNA 36.1,155–66.

Daly, A. and Nymoen, P., 2008, The Bøle ship,Skien, Norway – Research history, dendrochronology and Provenance, IJNA37.1, 153–70.

De Aranda y Antón, G., 1990, Los Bosques Flotantes: historia de un roble del siglo XVIII, Colección Técnica, Ministerio deAgricultura, Pesca y Alimentación, ICONA, Madrid.

De Aranda y Antón, G., 1999, El camino del hacha: la selvicultura, industria y sociedad. Visión histórica, Ministerio de MedioAmbiente, Organismo Autónomo de Parques Nacionales.

De Vries J. and Van der Woude A. M., 1995, Nederland 1500–1815. De eerste ronde van moderne economische groei, Amsterdam.Delmás, M. and Van den Berselaar, H., 2009, ‘Nederlands’ hout op drift. Over houthandelsroutes en de herkomst van hout van

de Late Middeleeuwen tot in de 18de eeuw, Vitruvius 2.6, 12–18.Devy-Vareta, N., 1986, Para uma geografia histórica da floresta portuguesa: do declínio das matas medievais à política florestal

do Renascimento (XV–XVI), Revista da Faculdade de Letras-Geografia 1.1, Porto, 5–37.Dodds J. and Moore J., 1984, Building the wooden fighting ship. Chatham, London.Eckstein D. and Bauch J., 1969, Beitrag zu Rationalisierung eines dendrochronologischen Verfahrens und zu Analyse seiner

Aussagesicherheit. Forstwissenschaftliches Centralblatt 88, 230–50.



Eckstein J., Leuschner H. H., Bauerochse A., and Sass-Klaassen U., 2009, Subfossil bog-pine horizons document climate andecosystem changes during the Mid-Holocene, Dendrochronologia 27.2/2, 129–46.

Feuillat, F., Dupouey, J. L., Sciama, D., and Keller, R., 1997, A new attempt at discrimination between Quercus petraea andQuercus robur based on wood anatomy, Canadian Journal of Forest Research 27, 343–51.

Fritts, H. C., 1976, Tree rings and climate, Academic Press, London.García Esteban, L., Guindeo Casasús, A., Peraza Oramas, C., and De Palacios de Palacios, P., 2003, La madera y su anatomía,

Fundación Conde del Valle de Salazar y Mundi-Prensa, AITIM, Madrid.Giachi, G., Lazzeri, S., Mariotti Lippi, M., Macchioni, N., and Paci, S., 2003, The wood of ‘C’ and ‘F’ Roman ships found in

the ancient harbour of Pisa (Tuscany, Italy): the utilization of different timbers and the probable geographical area whichsupplied them. Journal of Cultural Heritage 4, 269–83.

Grenier, R., Ringer, R. J., and Bernier, M. (eds), 2007, The Underwater Archaeology of Red Bay: Basque Shipbuilding andWhaling in the 16th Century. Ottowa.

Guibal, F. and Pomey, P., 2003, Timber Supply and Ancient Naval Architecture, in C. Beltrame (ed.), Boats,Ships and Shipyards Proceedings of the Ninth International Symposium on Boat and Ship Archaeology, 12/2000, 35–48,Venice.

Haneca K., Boeren, I., Van Acker, J., and Beeckman H., 2006, Dendrochronology in suboptimal conditions: tree-rings ofmedieval oak from Flanders (Belgium) as dating tool and archives of past forest management, Vegetation History andArchaeobotany 15.2, 137–44.

Hillam, J., 1979, Tree-rings and archaeology: some problems explained, Journal of Archaeological Science 6, 271–78.Klein, P. and Esteves, L., 2001, Dendrochronological analyses in Portuguese panel paintings, in R. van Schoute and H.

Verougstraete (eds.), Colloque XIII Dessin sous-jacent et la technologie dans la peinture: la peinture et le laboratoire, procédés,méthodologie, applications, Bruges, 15–17 septembre 1999, Uitgeverij Peeters, Leuven, 213–20, Bruges.

Locatelli, C., Lavier, C., Pousset, D., 2011, Synopsis des chantiers de bois de la cathédrale d’Auxerre depuis 1235, in C. Sapin(ed), Saint-Etienne d’Auxerre, la seconde vie d’une cathédrale, Actes de colloque d’Auxerre (27–28 Sept. 2007), Auxerre, Centred’études médiévales Saint-Germain d’Auxerre, Editions Picard, 177–81.

Loureiro, V., 2011, L’Épave Arade 1: l’influence des chantiers régionaux sur la tradition de construction navale ibéro-atlantique, Unpublished doctoral dissertation, Université de Paris I – Panthéon Sorbonne.

Loureiro, V. and Alves, J. G., 2005, O navio Arade1. Relatório da campanha arqueológica de 2004, Trabalhos do CNANS 28,Lisbon: Centro Nacional de Arqueologia Náutica e Subaquática.

Loureiro, V. and Alves, J. G., 2006, O navio Arade1. Relatório da campanha arqueológica de 2005, Trabalhos do CNANS 32,Lisbon: Centro Nacional de Arqueologia Náutica e Subaquática.

Loureiro, V. and Alves, J. G., 2008, The Arade 1 Shipwreck: Preliminary Results of the 2004 and 2005 Field Seasons, IJNA 37.2,273–82.

Mitchels, J. W., 1973, Dating methods in archaeology, Studies in Archaeology, Seminary Press, New York.Murmanis, L., 1975, Formation of tyloses in felled Quercus rubra L. Wood Science and Technology, Quercus rubra L. 9.1, 3–14.Orella Unzué, J. L., 2003, Comerciantes vascos en Normandía, Flandes y La Hansa, 1452–1526. Itsas memoria: revista de

estudios marítimos del País Vasco 4, 65–114.Otte Sander, E., 2008, Sevilla, siglo XVI: Materiales para su historia económica, Herederos de Enrique Otte Sander. Centro de

Estudios Andaluces. Consejería de la Presidencia, Junta de Andalucía.Pilcher, J. R., 1987, A 700 year dating chronology for northern France. Applications of tree-ring studies. Current research in

dendrochronology and related subjects. BAR International Series 333, 127–39.Queiroz, P. F., Oliveira, H., and Pereira, T., 2005, Identificação de um conjunto de madeiras provenientes da estrutura do navio

Arade 1, Trabalhos do CIPA 91. Lisbon.Rieth, E., 1989, Le clos des galées de Rouen, lieu de construction navale à clin et à carvel (12-93-1419), Medieval Ships and the

Birth of Technological Societies 1, 71–7.Rieth, E., Rodrigues, P., Alves, F., 2004, Relatório da campanha de 2003 de desmontagem e recuperação da parte exposta do

navio Arade I, Trabalhos do CNANS 19, Lisbon: Centro Nacional de Arqueologia Náutica e Subaquática.Schweingruber, F. H., 1990, Anatomy of European woods. Haupt Verlag, Bern, Stuttgart.Schweingruber, F. H., 1993, Trees and wood in dendrochronology. Springer Series in Wood Science. Springer Verlag, Heidel-

beerg.Schweingruber, F. H., 1996, Tree rings and Environment. Dendroecology. Paul Haupt Verlag, Berne.Sousa, V. B., Leal, S., Quilhó, T., and Pereira, H., 2009, Characterization of cork oak (Quercus suber) Wood anatomy. IAWA

Journal 30.2, 149–61.Tutin, T. G., Heywood, V. H., Burges, N. A., Valentine, D. H., Walters, S. M., and Webb, D. A., 2001. Flora Europaea.

Cambridge University Press, Cambridge.Tyers, I., 1997. Dendro for Windows Program Guide. Archaeological Research and Consultancy at the University of Sheffield,

Report 340.Van Leeuwaarden, W. and Queiroz, P. F., 2002, Identificação de um conjunto de madeiras provenientes da estrutura dos navios

recuperados no rio Arade, Trabalhos do CIPA 35, Lisboa.Wazny, T. and Eckstein, D., 1987, Der Holzhandel von Danzig/Gdansk -Geschichte, Umfang und Reichweite, Holz als

Roh-Werkstoff 45, 509–13.Wicha, S., Guibal, F., and Medail, F., 2003, Archaeobotanical characterization of three ancient Mediterranean shipwrecks. In

Fouache E. (ed.), The Mediterranean world environment and history, Elsevier, 233–37.



Documents

Dendrochronological dating and provenancing of timbers from the Arade 1 shipwreck, Portugal