The Cais do Sodré Shipwreck, Lisbon, Portugal

The International Journal of Nautical Archaeology (2011) 40.2: 328–343 doi: 10.1111/j.1095-9270.2011.00313.x

The Cais do Sodré Shipwreck, Lisbon, Portugal Filipe Castro, Kotaro Yamafune, Coral Eginton and Thomas Derryberry Center for Maritime Archaeology and Conservation, Department of Anthropology, Texas A&M University, College Station, TX 77843–4352, USA The timber remains of a shipwreck, probably dating to the late-15th or early-16th century and found in 1995 at Cais do Sodré, Lisbon, Portugal, during the construction of a subway station, are described and analysed. © 2011 The Authors Key words: Cais do Sodré Shipwreck, Portugal, 16th-century shipbuilding.

T

he Cais do Sodré shipwreck was found in April 1995 during excavation works for the construction of a new underground station near downtown Lisbon, Portugal, which became the terminus of the line Caravela—named after this find. Almost devoid of artefacts and showing few signs of wear on the outer surface of the keel and hull-planking, the remains of this ship were lying horizontally at a depth of c.6.5 m below the water-table, listing 14° to starboard. The wreck, which was presumably situated on an ancient river-bed c.120 m offshore from the Lisbon waterfront, is thought to have sunk around 1500 AD. The area where the ship was lying was eventually covered by 19th-century landfill (Fig. 1). The orientation of a breast-hook and the remains of a whipstaff suggest that the ship’s bow pointed north, in the direction of the shore. As found, the hull appeared to have been heavily damaged by the digging equipment. The upper sections of the structure were broken and showed no signs of exposure to natural marineenvironment erosion. It is not clear whether this ship sank at a shallow depth and was subsequently salvaged, or whether it was abandoned after an accident. The hull-planking showed a few obvious signs of repairs, although there was little evidence of marine worm attacks (Teredo or Limnoria) on the outer surfaces of the timbers. Two lead sheathing fragments are mentioned in an internal memorandum dated 3 September 1996, together with a musket ball and a small collection of artefacts, bone fragments, and two sheave blocks, one of which was complete. Since the artefacts were not positioned within the volume excavated (c.100 ¥ 24 ¥ 6.5 m) it is impossible to say whether they were associated with the ship remains (Fig. 2). One small iron gun and three anchors were also found in the same area, although it is not known whether they too are associated (Castro and Yamafune, 2010).

Analysis of the vessel was entrusted to archaeologist Paulo Jorge Rodrigues, who died unexpectedly in November 2008, before completing the work. This paper is a tribute to him. Unfortunately, Rodrigues was not present at the time of the find, and before the bulldozer operator realized he had hit a coherent wooden structure of archaeological importance, the central frames had been torn apart, together with the keelson, which was part of the mast-step arrangement (Fig. 3). Lisbon’s subway company, Metropolitano de Lisboa, reported the find to the cultural heritage agency in charge of archaeological supervision and excavation, then Instituto Português do Património Arquitectónico e Arqueológico (IPPAR). After cleaning and tagging the structure, the IPPAR team produced a site-plan with the help of a total station supplied by the contractor. Sections were taken every 50 cm, after which the hull-structure was disassembled, packed and delivered to IPPAR (Rodrigues, 1995). Another two hull-plans were produced during the disassembly phase. One was made after the removal of the remaining portions of the keelson, ceiling planking, stringers, and breast-hook, and shows the exposed framing system. A third plan shows the outer planking after the removal of the frames. Carpenter-marks were observed during the disassembly process but not thoroughly recorded, since the archaeologists planned to record each individual timber in greater detail during a future analysis stage. Unfortunately, IPPAR showed little interest in this shipwreck and stored the timbers in a warehouse without proper care for over six months, during which they dried and warped irreversibly. Rodrigues did not start the recording of the timbers until a year after the find, when IPPAR handed the timbers, already distorted, to Instituto Português de Arqueologia (IPA), a newly created organisation in charge of the country’s archaeological heritage.

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society. Published by Blackwell Publishing Ltd. 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.

F. CASTRO ET AL.: THE CAIS DO SODRÉ SHIPWRECK, LISBON, PORTUGAL

Figure 1. Location of the site. (maps after Google Earth)

Two samples of timber were radiocarbon-dated in Portugal at the time of the find, suggesting a mid-tolate-15th-century origin for this shipwreck (Rodrigues, 2002). Another two samples were collected in April 2010 from the outer rings of a floor-timber and a hullplank, and radiocarbon-dated in the USA, yielding dates in the mid-15th century and early-16th century (Table 1). The preliminary results of this project were presented at a conference on Iberian ships in Lisbon in 1998 (Rodrigues et al., 2001), and the importance of this shipwreck justified the organization of two Texas A&M University field-schools in Lisbon, in the summers of 2001 and 2002. With the support of Centro Nacional de Arqueologia Náutica e Subaquática (CNANS/IPA), Texas A&M University Nautical Archaeology Program (NAP) students recorded all the ship’s floor-timbers at 1:1 and 1:10 scales (Castro, 2001; 2002) and in 2002 the construction of a wooden model of the shipwreck was commenced. In 2002 Paulo Jorge Rodrigues finished his maîtrise at Sorbonne I University, under the orientation of Dr Eric Rieth (Rodrigues, 2002). The Texas A&M fieldschools in Lisbon were halted that year, and Rodrigues’s poor health and his eventual leave from IPA greatly slowed the rhythm of the project until his untimely death in 2008. In March 2010 the first author of this paper obtained permission to continue Rodrigues’s work, honour his memory, and publish this

interesting shipwreck in English. The first phase of research consisted of organizing, verifying the accuracy of, and presenting the existing primary data, while the second phase proposes an interpretation of the original research. This paper is a summary of the information available at both the Divisão de Arqueologia Náutica e Subaquática (DANS) at Instituto de Gestão do Património Arquitectónico e Arqueológico (IGESPAR) (the agency created in 2006 to replace IPPAR and IPA) and the Ship Reconstruction Laboratory at Texas A&M University. Following a visit to DANS and an assessment of the data available there, a methodology was established for each of the phases of this project. Four sources of primary data were identified: the drawings of the hull made with the total station before distortion; the 1:1 and 1:10 drawings of the distorted floor-timbers; Rodrigues’s sketches and measurements; and the first author’s own notes. An additional drawing of the top of the keel, in the area where the frames were removed by a bulldozer, was produced by DANS in July 2010, which allowed us to position the missing frames over the reconstructed model. We are grateful to Francisco Alves, head of DANS, and his team, namely Pedro Neves de Oliveira, João Coelho, and Miguel Aleluia, for their patience and constant support during the spring and summer of 2010.

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society

329

NAUTICAL ARCHAEOLOGY, 40.2

Figure 2. The shipwreck site in 1995, during the disassembly operation. (DANS)

Methodology This paper is primarily intended as an overview of the ship’s form, dimensions, and constructional features. As no complete timber catalogue was created during the excavation, the first step was to inventory all the timbers of this shipwreck and collect all the existing data on each timber (Tables 2–7). The second step was to produce sections of the planking plan, spaced 1 m apart, in order to understand and test the total-station data. A longitudinal section along each edge of the upper surface of the keel was produced as well, with the same objective. Once the data was organized, and the gaps and unclear details could be filled in and corrected in the original (total-station) drawings, a new set of 1:10-scale plans was produced and completed with one longitudinal section along a straight keel-axis on which the heights of the construction marks observed could be marked. The third and last step of this phase was to produce a short description of each timber-type, illustrated with the images and pictures available (Castro and Yamafune, 2010). We have planned a second phase, focusing on analysis of the construction-marks and on a study of the ship’s conception process. The shipwreck was pierced by more than 50 circular holes made during the process of laying the concrete foundations of the subway. In several places, these jet-holes caused tremendous destruction. When com330

pared to the in situ plans, the drawings of the floortimbers show evidence of only minor longitudinal shrinkage, even if the radial shrinkage and warping are considerable. The craftsmanship seemed excellent, even though the hull was badly damaged by the jetholes. The identification of the vessel’s type and function is difficult at this stage. The scantlings seem too light for its keel-length and there are pronounced kinks at both the bow and stern at a very low height above the keel; both of which are not common features in ocean-going ships. The heights of the entries (in Portuguese delgados da proa) and the runs (delgados da popa) are not proportional to the ship’s keel-length, and a first sketch of the ship’s lines suggests a wide bottom amidships with sharp, albeit low, bow and stern. Only future analysis will allow us to make educated guesses about the nature and geographical origin of this ship. This paper focuses on retrieving, organizing, and making sense of the existing data.

Hull remains The Cais do Sodré shipwreck was found lying on a roughly north-south orientation with its presumed bow pointing north. The surviving remains consisted of a section of the keel almost 24 m long, fragments of 37 frames, one apron, part of the keelson, two stringers, some ceiling planking, one breast-hook, one orlop

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society

F. CASTRO ET AL.: THE CAIS DO SODRÉ SHIPWRECK, LISBON, PORTUGAL

Figure 3. Hull-plan drawings: keelson, ceiling, stringers, breast-hook and whipstaff; frames; and planking. (F. Castro and K. Yamafune, 2010) Table 1. Radiocarbon dates of Cais do Sodré timbers Date

Sample reference

Sample nature

Calibrated date (2d)

1995 1995 2010 2010

Sac-1334 Sac-1335 Beta-279091 Beta-279090

Hull-plank Floor-timber Hull-plank Floor-timber C148

Cal AD 1424–1516 and 1590–1622 Cal AD 1424–1516 and 1590–1622 Cal AD 1520–1580* Cal AD 1420–1490

* and Cal AD 1630–1680, Cal AD 1770–1800, and AD 1940–1950.

beam, one mast-step buttress, one fragment of a stanchion, and part of a whipstaff. The hull-planking was preserved along the full length of the wreck. The planking maximum breadth preserved was c.4.9 m, and encompassed seven strakes to starboard and nine to port. Two stringers, one breast-hook and nine strakes of ceiling planking (five to port and four to starboard) were still in place, as well as the stump of a stanchion, with its tenon still inserted in a mortise cut into the upper face of the keelson. After a re-analysis of the primary data, to clarify any existing doubts, new 1:10 drawings were produced

(Fig. 3), and all timbers were numbered, referenced and measured (Fig. 4). Sections of the hull spaced 1 m apart were drafted from the total-station drawings, and a preliminary hull analysis was performed with the use of the 3D-modelling software Rhinoceros® (Figs 5–7). We based our analysis on Rodrigues’s drawings and notes, and kept the original timber designations whenever possible. However, changes were made after clarification of certain details in the original drawings, and the resulting nomenclature is a mix of Portuguese and English initials (‘O’ and ‘E’ stand for West (Oeste) and East (Este), ‘TC’ for tábua de casco

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society

331

NAUTICAL ARCHAEOLOGY, 40.2 Table 2. Scantlings of the Cais do Sodré shipwreck: ceiling (thickness 4–5 cm) Strakes North—Bow: port side Strake 1 Strake 2 North—Bow: starboard Strake 1 Strake 2 South—Stern: port side Strake 1 Strake 2 Strake 3 Strake 4

Strake 5

South—Stern: starboard Strake 1 Strake 2 Strake 3 Strake 4

Ref.

Length (m)

Sided (cm)

Wood species

O1.1N O1.2N O2.1N

2.00 1.32 3.30

38 24 30

Quercus faginea Quercus faginea Quercus faginea

E1.1N E2.1N

2.84 1.16

15 18

Quercus faginea Quercus faginea

O1.1S O1.2S O2.1S O2.2S O3.1S O3.2S O4.1S O4.2S O4.3S O4.4S O5.1S O5.2S O5.3S O5.4S

1.82 4.52 4.52 0.44 4.60 2.74 0.70 3.64 2.76 2.00 1.70 1.38 1.02 1.90

17 27 28 32 32 48 8–10 10–16 20 20 15 10 10 28

Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus faginea Pinus pinea Pinus pinea Pinus pinea Pinus pinea Quercus faginea Quercus faginea Quercus faginea Quercus faginea

E1.1S E2.1S E2.2S E3.1S E3.2S E4.1S

5.18 1.54 5.16 1.40 5.14 1.56

16–20 16–20 18–24 18 24 12

Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus faginea Pinus sylvestris

Table 3. Scantlings of the Cais do Sodré shipwreck: structural timbers Timber Keel

Keelson

Apron 1st Stringer

2nd Stringer Breast-hook Stanchion Orlop beam Buttress Whipstaff

332

Ref.

Length (m)

K1 K2 K3 K4 KS1 KS2 KS3 KS4 A SO1.1 SO1.2 SO1.3 SE1.1 SO2.1 SO2.2 SE2.1 B1 St1 Ob1 Bt1 Ws

8.48 3.39 7.57 2.97 1.30 4.88 2.60 4.00 2.20 2.80 0.65 8.75 4.40 4.60 4.80 2.25 1.70 0.54 1.16 1.30 0.75

Sided (cm)

Moulded (cm)

Wood species

25

27

27

26

22.5 18

15 17

18

17

Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus faginea Quercus rubor Quercus faginea Quercus faginea Quercus faginea Crataegus monogyna

23 10 18 20 Ø8

25–29 10 18 18 —

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society

F. CASTRO ET AL.: THE CAIS DO SODRÉ SHIPWRECK, LISBON, PORTUGAL Table 4. Scantlings of the Cais do Sodré shipwreck: hull planking (all planks Quercus faginea, 7–8 cm thick) Strakes

Ref.

Strake 1

TCE 1.1 TCE 1.2 TCE 1.3 TCE 1.4 TCE 2.1 TCE 2.2 TCE 2.3 TCE 2.4 TCE 2.4a TCE 3.1 TCE 3.2 TCE 3.3 TCE 3.4 TCE 3.5 TCE 4.1 TCE 4.2 TCE 4.3 TCE 5.1 TCE 5.2 TCE 5.3 TCE 5.4 TCE 6.1 TCE 6.2 TCE 7.1

Strake 2 Strake 3

Stealer Strake 4 Strake 5

Strake 6 Strake 7

Length (m) Starboard 5.45 8.00 5.70 6.80 6.25 2.20 0.95 8.40 0.95 3.30 2.65 6.10 2.67 3.40 4.35 7.70 0.85 1.17 2.15 3.40 2.95 1.75 3.35 1.15

Sided (cm)

Strakes

Ref.

22 38 41 27 35 25 43 45 9 28 41 40 30 19 15 33 14 20 25 43 23 30 35 21

Strake 1

TCO 1.1 TCO 1.2 TCO 1.3 TCO 2.1 TCO 2.2 TCO 2.3 TCO 3.1 TCO 3.2 TCO 3.2a TCO 3.3 TCO 4.1 TCO 4.2 TCO 4.2a TCO 5.1 TCO 5.1a TCO 5.2 TCO 6.1 TCO 6.2 TCO 6.2a TCO 6.3 TCO 7.1 TCO 7.2 TCO 7.3 TCO 7.4

Strake 2 Strake 3 Repair Strake 4 Strake 5 Stealer Strake 6

Strake 7

Length (m) Port

Sided (cm)

11.20 5.40 4.80 9.30 7.80 5.30 16.50 4.90 0.71 1.70 10.50 12.80 2.08 4.40 2.20 6.20 2.00 8.75 2.20 8.70 2.65 1.15 11.55 2.00

30 33 18 45 45 29 39 40 6 15 45 52 25 35 10 22 22 48 8 39 35 30 32 33

Table 5. Framing pattern: filling floor timbers (south) Frame No. 77 78 79 80 81 82 83 84 85

Sided dimension

Space to next frame

Height over the keel

Height 1 m from centre

19 20 18 21 20 21 24 23 18

17 43 28 24 24 21 25 24 19

— 47 36 34 37 36 41 39 38

— 18 15 16 17 18 17 16 17

(hull-plank) and ‘S’ for strake). The approximate dimensions of the timbers preserved are indicated on Tables 2–4. There is a substantial amount of evidence suggesting that this ship was constructed following a method common in the Mediterranean and the Iberian Peninsula, in which the central frames are pre-designed and pre-erected to define the shape of a central portion of the hull, and the ends are fashioned by eye using ribbands (armadouras). The central frames showed a number of interesting features including constructionmarks cut on the faces of some of the floor-timbers (Fig. 8). These marks consisted of a sequential numbering of 18 frames on both sides of a single masterframe, a series of vertical lines marking the edges of the

keel (in Portuguese astilhas), and two sets of lines in the zone of the turn of the bilge. Since the original excavation notes are incomplete, remaining timber dimensions were taken from the original 1:10-scale drawings. There are doubts as to the interpretation of some of these drawings; namely, which dimensions pertain to the total length of some hull-planks, which at times seem implausibly long. Such are the cases of: planks TCE 1.1 and TCE 1.2, the seam between them is not clearly indicated in the original; plank TCO 3.1, which probably corresponds to two planks; planks TCO 4.1, 4.2, and 7.3, implausibly long. A second problem in the interpretation of the data is associated with the difficulty of finding the centre plane of symmetry of each frame. Without this

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society

333

NAUTICAL ARCHAEOLOGY, 40.2 Table 6. Framing pattern: pre-designed frames Frame No.

Roman Numeral

86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 Missing 140 141 142

Sided dimension

Space to next frame

Height over the keel

Height 1 m from centre

21 22 20 19 21 19 18 21 19 16 19 21 19 19 19 — 20 20 20

21 15 18 16 15 26 19 23 18 25 22 15 18 24 — — 30 11 14

40 35 32 35 30 33 34 31 31 30 31 30 30 31 32 — 38 38 36

19 18 18 18 18 18 21 23 19 18 19 19 20 18 20 — 20 20 17

(I)IIVX XVII* IVX XV* IIIIX XIII XII XI Not visible V(IIII) VIII* IIV IV V II(I)I III to XV Not visible IIV(X) VXIII

* Upside down.

Table 7. Framing pattern: filling floor timbers (north) Frame No. 143 144 145 146 147 148 150 152 154 156 158

Sided dimension

Space to next frame

Height over the keel

Height 1 m from centre

24 22 17 17 22 20 20 20 20 20 23

23 23 28 21 23 13 19 21 9 34 —

34 34 42 32 46 37 27 43 43* 38* 33*

18 15 21 17 16 14 17 19 17 14 12

* Over the apron.

plane it is impossible to determine the exact co-ordinates of the construction marks. Keel The keel was preserved along nearly the entire distance between the concrete walls of the subway station. It presented the same rectangular transverse section throughout its entire length, 25 cm sided and 27 cm moulded. The ship’s bow and stern were cut during the excavation process and the ends of the hull were badly disturbed. At the bow, the keel was twisted to port and the entire structure was incoherent near the wall. As indicated in Table 2, the keel was composed of four timbers of different lengths. Each section butted against the next without any scarf. A water-stop dowel was inserted under the rabbet line in all three preserved 334

connections (Fig. 9). This solution is illustrated in the anonymous text Traité de Construction de Galères, which dates to the late-17th century (Rodrigues, 2002: 15), but has no parallel in other published Iberian ships. As Rodrigues noted, a butt scarf was also found on the connection between the Culip VI keel and stempost. The keel’s lower face showed signs of use, making it unlikely that there was ever a shoe underneath. Apron Part of a badly broken apron was preserved in the bow area. It measured 225 mm sided and 150 mm moulded and the upper surface of its forward portion curved upwards with a radius of c.80–100 cm. The apron’s upper surface was notched on its sides to fit the feet of ‘V-shaped’ frames. The notches were c.5 cm deep and

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society

F. CASTRO ET AL.: THE CAIS DO SODRÉ SHIPWRECK, LISBON, PORTUGAL

Figure 4. Timber references: keelson, ceiling, stringers, breast-hook and whipstaff; frames; and planking. (F. Castro and K. Yamafune, 2010)

Figure 5. Hull-planking rendering in Rhinoceros®: longitudinal view. (C. Eginton)

Figure 6. Hull-planking rendering in Rhinoceros®: plan. (C. Eginton)

5 cm wide, and as long as the sided dimension of the frames they received. The apron was bolted to the keel and possibly to the stempost as well, although its upper section was not preserved. No scale drawings were made of this timber, but its curvature was reconstructed from the total-station co-ordinates in the respective reconstruction. The longitudinal

representation in Fig. 10 is conjectural, based on the data available for the timber’s upper surface. Keelson The keelson was partially torn apart by the bulldozer operator. Four sections were recorded, three on the site-plan made with the total station, and the fourth

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society

335

NAUTICAL ARCHAEOLOGY, 40.2

later, in 1997, by the first author. The keelson had a constant transverse section (27 cm sided ¥ 26 cm moulded) and was notched underneath, to fit over the floor-timbers. Fragments KS1 and KS2 were still in place and connected through a horizontal hooked scarf. Section KS3 was positioned during the reconstruction process by comparing the distances between fastening-holes. The only likely position is indicated on Fig. 11, although at this point it is impossible to establish whether its scarf was pointing to the bow or the stern. The keelson was fastened to the keel with iron bolts c.30 mm in diameter, presumably inserted from underneath, at irregular intervals, generally every two or three frames.

Figure 7. Hull-planking rendering in Rhinoceros®: perspective. (C. Eginton)

Frames A total of 43 frames were preserved, all in their original positions, 19 to the north, in the bow area, and 24 to the south, in the stern area (Tables 5–7). Each of the floor-timbers was drafted at 1:1 and 1:10 scale in 2001 and 2002. The preserved frames fall into two clear categories: pre-designed frames (cavernas graminhadas) and filling floor-timbers (enchimentos). Each predesigned frame was composed of one floor-timber and two futtocks, and was fastened to the keel with two iron spikes, inserted through recesses cut on the forward face of the stern frames and on the aft face of

Figure 8. Floor-timber C87 showing the Roman numerals, the marks of the sides of the keel, and the horizontal and vertical lines. (F. Castro) 336

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society

F. CASTRO ET AL.: THE CAIS DO SODRÉ SHIPWRECK, LISBON, PORTUGAL

Figure 9. Keel joint. (K. Humdinger)

Figure 10. Detail of the horizontal projection of the planking and reconstructed longitudinal view of the hull showing the apron with the frames sitting on it. (F. Castro and K. Yamafune)

the bow frames. Each futtock was fastened to its floortimber with a dovetail scarf and three iron spikes, always inserted from the floor-timber side. The dovetails were always wider at the lower side, which measured between 17 and 31 cm. The dovetail tops measured between 11 and 23 cm. Dovetails were salient on all floor-timbers except C93, where the only dovetail that survived, on the port side, was recessed. Dovetail thicknesses varied between 1.5 and 2 cm, or around the value of the 16th century dedo (1.8 cm). These pre-designed frames were numbered with Roman numerals from I to XVIII before and after a single master-frame, which was torn apart

by the bulldozers and thrown away before the archaeologists became involved with the shipwreck. The master-frame was fastened to the keel with one iron spike and two iron bolts. The pre-designed frames showed three types of construction-marks (Fig. 8): two vertical lines marking the foot of the frame, where it sits on the keel (these lines are irregular and often do not allow us clearly to establish the design vertical axis of the frame); one or two horizontal lines (in some cases not horizontal at all) which may have marked the base of the mould from which all floor-timbers seem to have been cut; and one or two vertical marks near the turn of the bilge arc, to the outside of the horizontal line, in reality diagonal, whose significance is not clear at this point. The filling floor-timbers were fastened to the keel with a single iron spike, inserted in a recess cut for that purpose, and were not fastened to their respective futtocks. These timbers were not numbered. However, six timbers (nos 80, 82, 83, 85, 150, 152) showed evidence of vestigial diagonal lines which may be related to the construction process. Timbers C78 and C81 had one arm scarfed into their lower section, presumably for lack of suitable ‘Y-shaped’ timbers. Almost all floortimbers (C89 to C100, C140 to C145, and C148) had one limber-hole positioned on the port side, between 20 and 45 cm from the central axis of the ship. Floortimbers C90 and C96 had two limber-holes each. The positions of the missing frames were not recorded, based on the fastening-marks of the hullplanking during the initial excavation. At a later date Rodrigues made a 1:10-scale drawing of the upper face of keel section K2, and in July 2010 a team of students under the orientation of Francisco Alves made a 1:1scale drawing of the upper face of keel section K3. From these drawings, the positions of the missing frames were easily deduced by interpreting the clear fastening pattern illustrated on both drawings. The distances between fastening-holes are indicated in Table 6, together with the sided and moulded dimensions of each frame, the first measured in the more accurate total-station drawing, and the second in the 1:10 drawings made by the Texas A&M teams in 2001 and 2002 from the dried and warped timbers. The scantlings presented below are therefore average

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society

337

NAUTICAL ARCHAEOLOGY, 40.2

Figure 11. Longitudinal section of the ship remains showing the preserved portions of the keelson and the positions of the bolts that fastened the keel and the keelson through the floor-timbers. (K. Yamafune and F. Castro)

values, measured on the dried timbers, after considerable radial shrinkage and warping had occurred. Room and space are indicated as measured in the totalstation drawings, which suggest that the distance between the concrete walls was not 24 m, but rather 23.95 m. The positions of the construction-marks were measured by Rodrigues on the dried and warped timbers. When measured again in the ShipLab, from the 1:10 drawings of the frames, there were a number of differences, always associated with the difficulty of establishing their vertical axis. The values presented below are those measured by Rodrigues directly from the frames (2002: 37). Further analysis by Mauro Bondioli and Mariangela Nicolardi has allowed us to refine these values and will be presented in a future publication. The futtocks were never recorded at 1:1 scale, but as none was preserved beyond the tips of the floortimbers, this lack of data is not particularly important at this point. Planking The planking was 7–8 cm thick. All boards were cut from Portuguese oak (Quercus faginea), a tree quite common in the western Mediterranean (the Iberian peninsula and the north of Africa). No caulking arrangement was reported during disassembly. It is not known whether the two lead-sheathing fragments mentioned in the 1996 memorandum were found in association with the hull-planking, or whether there were any caulking remains between the planks. Nine strakes of hull were preserved to port and seven to starboard. Drop strakes and stealers were used to achieve the sharp dead-rise at the bow and stern (Fig. 4; Table 4). Most planks were cut from large oaks, presenting a maximum length of c.8 m and average widths between 20 and 50 cm. Each 338

plank was fastened to the frames with two spikes per frame and three near the hoods. Wider planks had 3 spikes per frame and four spikes near the hoods. A conversion study was not performed on the hullplanks, and at this point it is difficult to envisage such a study, given that most of the planks’ labels fell off or are illegible. The garboards had the same thickness as the rest of the hull and were chamfered to fit the keel rabbets. An analysis of the total-station data showed that the hull was heavily hogged in its undisturbed condition, and that the planking opened and sank after the removal of the ceiling and framing (Fig. 12). Once we manage to reconstruct the hull shape in a plausible configuration, the extent to which the hull-planking was preserved and recorded should allow us to reconstruct the shape of the stern- and stemposts, by extending each plank in a 3D model. Ceiling The ceiling planking was made of 4–5-cm boards of Portuguese oak (Quercus faginea) and preserved near the bow and the stern of the wreck (Fig. 4; Table 2). In the bow portion of the vessel, only two strakes on each side of the keelson survived, and they were too damaged to allow for proper analysis. In the stern area the strakes were well preserved, with five port and four starboard strakes remaining. Their condition suggested that good and competent carpenters laid them carefully. The maximum length of the surviving planks was over 4 m, and widths varied between 10 and 48 cm. Breast-hook One breast-hook was preserved in the northern portion of the hull (Fig. 13). It was the only surviving timber which was cut from an English oak (Quercus robur).

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society

F. CASTRO ET AL.: THE CAIS DO SODRÉ SHIPWRECK, LISBON, PORTUGAL

Figure 12. Planking shape after the removal of the ceiling and framing structures. (C. Eginton)

Figure 13. Breast-hook in situ. (F. Castro and K. Yamafune)

The breast-hook was roughly carved from a Y-shaped timber 1.7 m long, and had a maximum sided dimension of 60 cm with moulded dimensions ranging from 30 cm at the forwardmost section to 45 cm in the central zone. The breast-hook connected to the lower stringers through hooked scarves c.42 cm long and was spiked to the frames.

section (Fig. 4; Table 3). The stringers were 18 cm sided and 17 cm moulded and connected through horizontal hooked scarfs c.60 cm long. According to Rodrigues’s notes, the stringers were spiked and bolted to the frames, with two spikes per frame, but the rhythm of the bolting is not indicated (2002: 23).

Stringers Two stringers ran along the bottom of the ship and along the turn of the bilge, separated by the only pine planks identified in the ship. Only a small portion of the lower stringer survived in the northern portion of the hull, but both stringers survived in the southern

Orlop beam One beam placed low in the hull can be seen in the total-station site-plan and in some of the pictures (Fig. 14). It is concave and rests on a notch on the upper stringer. At this point, we have not been able to find a drawing of this timber.

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society

339

NAUTICAL ARCHAEOLOGY, 40.2

Figure 14. Possible orlop beam in situ, standing on the upper stringer. (Filipe Castro, 1995)

can be seen on a small rectangular ceiling-board, which defined a rectangular hole 25 ¥ 20 cm where a structure seems to have been placed. If this was the location of the mainmast-step, it is possible that these notches were related to the pump-sump structure. Stanchion A stanchion with a square section of 10 cm each side was partially preserved along 54 cm, still inserted in a mortise, over frame C144 (Fig. 16). Whipstaff The lower portion of a whipstaff was found lying on the ceiling-planking over frame C88. It has not yet been recorded at 1:1 scale. Cut from a hawthorn tree (Crataegus monogyna), it was preserved along 75 cm of its length, with a diameter of 8 cm. The lower ring is 28 cm in diameter and the hole is 14 cm. The thickness of the lower ring is also about 8 cm (Fig. 17).

Preliminary analysis Figure 15. Floor-timber C98, buttress, and the possible area of the mast-step and pump sump. (F. Castro and K. Yamafune, 2010)

Buttress and pump sump? One timber, found lying (not fastened) over floortimber C98, was thought to be a buttress (Fig. 15). Both the buttress and floor-timber C98 were notched on their stern faces (C98 on both sides of the keel) with notches 5 cm deep and 25 cm wide, at 40 cm to port and 50 cm to starboard of the keel axis. A similar cut 340

The preliminary study of the Cais do Sodré hullremains is rather puzzling. Summarizing, the scantlings seem light for an ocean-going ship with a keel-length of over 24 m; the keel scarfs have no known parallels in ocean-going ships; the bow and stern portions are missing; the pronounced outward kinks at the bow and stern frames make it difficult to imagine the depth-ofhold of this ship; and the whipstaff suggests a ship with more than one deck. It is difficult to theorize about the site-formation process because the upper portion of the frames and planking were destroyed by construction equipment, and so was the mast-step arrangement and whatever bulkheads or other structures remained on

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society

F. CASTRO ET AL.: THE CAIS DO SODRÉ SHIPWRECK, LISBON, PORTUGAL

Figure 16. Stanchion fragment in situ. (Paulo Jorge Rodrigues, 1995)

Figure 17. Whipstaff fragment in situ. (Paulo Jorge Rodrigues, 1995)

site. The lack of a substantial amount of ballast (the only possible ballast was a small amount of fine gravel in the mud deposited over the ceiling-planking) suggests that it was a derelict vessel, but the depth at which it seems to have been abandoned (c.6.5 m) seems too far away from the low-tide beach line. We have plotted two of the lines defined by the marks on the frames, and got a fair curve (Fig. 18) but no clear units of contemporary measure (neither dedos, palmos de vara or de goa, nor codos castellanos or cantabricos) seem to fit the model. After measuring a flat amidships of 4.16 m on a 1:20 drawing, we compared

the measurements of the construction-marks taken by Rodrigues (Table 8) with those obtained for a mezzaluna, two types of incremental triangles, and a Fibonacci series. The results are insufficient to formulate any conclusions, although the curve obtained is rather smooth, closer to a mezzaluna than the other, sharper curves. The values observed are unusual as well: the height of the construction-marks on the fore and aft tailframes, which are normally different, seem to be the same in this vessel at near 65 cm (in this case almost 3 palmos de vara), while the values of the bottom width

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society

341

NAUTICAL ARCHAEOLOGY, 40.2

Figure 18. Curves defined by the horizontal and vertical lines on the floor-timbers, as measured by P. Rodrigues. (F. Castro and K. Yamafune) Table 8. Framing pattern: pre-designed frames, Paulo Rodrigues’s co-ordinates Frame No.

Roman Numeral

Horizontal line x

Horizontal line y

86 (I)IIVX 117 87 XVII* 122/129 88 IVX 130 89 XV* — 90 IIIIX 150 91 XIII 158 92 XII — 93 XI — 94 Not visible — 95 V(IIII) 183 96 VIII* — 97 IIV 198 98 IV — 99 V 201 100 II(I)I 205 Central 19 frames missing: 3 abaft, one master frame, and 15 forward 140 Not visible — 141 IIV(X) — 142 VXIII 107

Vertical line x

Vertical line y

60 61/60 55 — 52 54 — — — 47 — 44 — 45 44

— 134 — — 169 175 — — — — — — — 226 224

— 63 — — 62 60 — — — — — — — 51 50

— — 65

— —

— —

* Upside down.

of the fore and aft tail-frames, which are normally equal, are different at c.2.34 m forward and 2.14 m aft. When measured along the line presented on Fig. 18, the maximum flat amidships seems to be 4.16 m, and the height of the turn of the bilge point close to 42 cm. These values give a total rising of [65 - 42 =] 23 cm fore and aft, and a total narrowing of [4.16 - 2.14 =] 2.02 m abaft, and [4.16 - 2.34 =] 1.82 m forward. If the rising forward seems plausible, around one palmo de vara, one would expect the rising abaft to be slightly higher, perhaps 1.5 times the value of the rising forward. The narrowing of the bottom is within a plausible range of values for the Iberian tradition, although it is not common that it is more pronounced abaft. The 342

bottom of the ship looks peculiar for this reason, as well as for the pronounced kink of its bow and stern frames outwards, which occurs very low considering the keel-length preserved. When plotted in a linesdrawing, these kinks are aligned in a perfect diagonal, more inclined at the stern than at the bow, and remind us of the Calvi 1 shipwreck stern (Villié, 1989; 1990; 1991). Only further research will answer these questions. We hope that the publication of the archaeological raw data—in preparation—will arouse curiosity other specialists in the interpretation and reconstruction of this interesting hull. We have already started the reconstruction of this shipwreck, but

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society

F. CASTRO ET AL.: THE CAIS DO SODRÉ SHIPWRECK, LISBON, PORTUGAL

Figure 19. Tri-dimensional model of the ship’s frames mounted on a straight keel. (T. Derryberry)

the work ahead seems lengthy and will depend ultimately on the number of students interested in this project.

Further analysis To extract the maximum amount of information from such a shipwreck site requires the construction of several models—educated guesses—and the testing of

their plausibility. We are planning to create a number of lines-drawings which will be tested against known ship-types. A wooden 1:20 scale 3D model is under construction, and two independent informatics 3D models have been started (for example Fig. 19). The next step in this project is the reassessment of the construction-mark co-ordinates and a tentative interpretation of them in light of the known conception recipes for that period.

Acknowledgements This project was only possible through the cooperation of IGESPAR. We are extremely grateful to IGESPAR president, Dr Gonçalo Couceiro, and DANS director Francisco Alves, whose team, namely Pedro Neves de Oliveira, João Coelho and Miguel Aleluia, provided us with constant and patient support during the spring and summer of 2010. This project is theirs.

References Castro, F., 2001, Relatório dos trabalhos de registo arqueográfico das madeiras do navio do Cais do Sodré, Verão de 2001, unpublished report on file in IPA/CNANS’ library. Castro, F., 2002, The Cais do Sodré Ship Frames—2002 Field Season—ShipLab Report 4, unpublished report, on file in IPA/CNANS’ library and Nautical Archaeological Program Library, Texas A&M University. Castro, F. and Yamafune, K., 2010, The Cais do Sodré Ship—ShipLab Report 13, unpublished report on file in IGESPAR/ DANS’ library and Nautical Archaeological Program Library, Texas A&M University. Rodrigues, P., 1995, Relatório Preliminar dos trabalhos de desobstrução e registo arqueográfico dos restos do navio encontrado no Cais do Sodré, nas obras do Metropolitano de Lisboa. Lisbon. Rodrigues, P., 2002, Étude de la charpente transversale du navire de Cais do Sodré de la 2ème moitié du XVe siècle/début du XVIe, masters dissertation, Université de Paris I—Sorbonne. Rodrigues, P., Alves, F., Rieth, E. and Castro, F., 2001, L’épave d’un navire de la moitié du XV.ème siècle/début du XVI.ème, trouvée au Cais do Sodré (Lisbonne). Note Preliminary, in F. Alves (ed.), Proceedings of the International Symposium, Archaeology of Medieval and Modern Ships of Iberian-Atlantic Tradition, 347–80. Lisbon. Villié, P., 1989, L’épave Calvi I, Cahiers d’Archéologie Subaquatique 8, 19–56. Villié, P., 1990, L’épave Calvi I, Cahiers d’Archéologie Subaquatique 9, 83–130. Villié, P., 1991, L’épave Calvi I, Cahiers d’Archéologie Subaquatique 10, 69–108.

© 2011 The Authors. International Journal of Nautical Archaeology © 2011 The Nautical Archaeology Society

343