Saint Gervais 3, mid-2nd century AD
Country: France
Place: Gulf of Fos (Bouches-du-Rhône), about 400 m from the shore, at a depth of 4 m, in the submerged area of the Roman port of Fossae Marianae
Coordinates (approximate): 43.42, 4.92
Type: Roman Imperial merchant ship of the western Mediterranean type
Identified: No
Dated: Mid-2nd century AD, based on the tituli picti on the Dressel 20 amphorae.
Beam: 7.50 m (master frame).
Keel Length: 8 m preserved (12.50 m of backbone).
Length Overall: Reconstructed at 17 m (Roman).
Number of Masts: Two: a main mast and a foremast (artemon).
History of the wreck
The Saint Gervais 3 wreck lies in the Gulf of Fos at a depth of 4 m and about 400 m from the shore (Liou and Gassend 1990), in an area with former land surfaces (cemeteries), now submerged, belonging to the Roman port of Fossae Marianae. Everything suggests that the wreck occurred around the middle of the 2nd century AD, thanks to the dating of the tituli picti preserved on a small group of Dressel 20 amphorae. The pottery of this cargo was produced, filled and shipped from an area far upstream of Hispalis, at Genil–Malpica, in the area of Écija (Astigi), and not from the mouth of the Guadalquivir itself (Liou and Gassend 1990: 259), which opens the way to relating the characteristics of the ship, the cargo and the palaeotopography of the production centre. Much of its cargo was salvaged right after the wreck, which explains why the amount of material recovered during the rescue excavation of 1979 was not large. The hull remains, however, were in perfect condition. Four excavation campaigns, with the participation of Gassend, between 1983 and 1986 allowed complete recording of the hull. This wreck represents the Imperial Roman type of the western Mediterranean (Pomey 1998a: 68–69; Pomey and Rieth 2005: 166–171).
Some of the bow floor timbers and half-frames were displaced from their original position and compressed against each other, indicating that the ship had suffered a violent collision at the bow, which caused it to sink (Liou and Gassend 1990: 219). Several marks reading QMF were found, which could be the initials of the tria nomina of the shipwright, as a signature. The keelson bears a series of punched letters reading CANTIORUM six times and RRFG at least four times. Cantiorum is the genitive plural of the family name Cantius, of Celtic origin, found mainly in north-eastern Italy (Aquileia) (Liou and Gassend 1990: 245), so it seems that a member of this family took part in the building process.
Description of the site
The hull, with its keel in situ, is preserved over 14.70 m of length and 6.80 m of beam (Figures 1 and 2). It rests on a practically flat seabed, and both sides survive fairly symmetrically, with a maximum width of 3.70 m on the starboard side and 3.10 m on the port side.
Figure 1. Simplified plan of the Saint Gervais 3 (after Liou and Gassend 1990: 220).
Figure 2. Lines plan of the Saint Gervais 3 wreck, from the longitudinal and cross-sections (after Liou and Gassend 1990: 261).
Cargo
Dressel 20 amphorae with tituli picti, from the area of Écija (Astigi). Much of the cargo was salvaged in antiquity, right after the wreck.
Ballast
Not reported.
Anchors
Not reported.
Guns
Not applicable.
Iron concretions
Not reported. Copper bolts and nails were used in the construction.
Hull Remains
The stem and part of the forefoot joining it to the keel had disappeared (Figure 3). The hull was coated internally with pitch. The longitudinal section of the keel shows that it is quite strongly curved and did not have the deformations characteristic of a ship that had been in service for a long time (Figure 3). At the stern, the sternpost described a gentle, regular curve, unlike the stem, which had disappeared. The very pinched shape of floor timbers M152 and M153, which break the alternation with the half-frames, indicates that the bow closed abruptly and sharply, so the hull had a concave bow.
Keel, keelson and mast step
The keel (Pinus halepensis) is preserved over 8 m and is joined by a trait de Jupiter (rayo de Júpiter) scarf to the forefoot, which is 3.60 m long; through another scarf a further section of 0.90 m (the sternpost proper) is preserved (Figure 1), so that 12.50 m of the backbone survive in total (Figure 3). The section of the keel in the bow area was 18 cm high and up to 17 cm wide, reduced in fact to 14 cm at the upper face because of the bevelled rabbets on its upper edges. At frame M110 the sternpost is 22 cm high and 18 cm wide at its upper face and 15 cm at its lower face.
In all known wrecks the mast step also serves as the keelson, except in the case of the Saint Gervais 3. Normally, in the area of the mast mortise, the piece is higher than in the section serving as keelson. The mast step proper in the Saint Gervais 3 is a timber of Abies alba 4.80 m long, 38 cm wide and 21 cm high (Figure 4). Its forward end coincides with frame M145 and its after end with M127, the master frame or maximum beam of the ship, so it lies towards the forward part of the ship. The main mast is between 2.45 and 2.80 m from the forward end (Figure 4), between M136 and M135 and 2 m forward of the master frame or maximum beam, that is, at 3/8 of the length. The mast cavity is 30 cm long, 10 cm wide and 9 cm deep, straight towards the stern and rising with a rounded section towards the bow. The foot of the mast has left its impression on the upper face of the mast step, 30 cm long and 17.5 cm wide, between the two lateral and the front cavities.
In the mast step there are four oblique notches or mortises, inclined at 60° to the horizontal: one towards the bow, on the starboard side face and inclined forward, and of the other three, two are just aft of the mast on each side face and the last at the after end, on the port side face (inclined aft) (Figure 4). They held timbers partly fitted and fastened by three nails, related to the stanchions. One of these, more than 3 m long, was found with its fitting in the mast step. It was a very important stanchion, since it allowed the depth or height to the deck of the ship to be established.1
The mast step sits on the keelson and is joined to it by six tenons arranged in staggered pairs. The tenons are 18 cm long, 10 cm wide and 6 mm thick (Figure 4). The fastening is completed by a square block of 18 cm and 2 cm thick on the underside of the mast step, fitting into the corresponding mortise in the keelson. The keelson is an enormous timber, 10.50 m long (Pinus halepensis), with an average width of 47 cm and a height of 27 cm, tapering towards the ends (to 15 cm at the stern). Its upper face is completely flat and its underside has the same curvature (or rocker) as the keel. It begins at the first surviving floor timber, M155, and ends at M115, before the position of the bilge pump, 4 m from the end of the surviving hull. On the upper face of the keelson, 1.40 m from its start, there is a mortise for a second mast at the bow (artemon). It is 30 cm long, 10 cm wide and 10 cm deep, with the usual rounded profile rising towards the bow.
The keelson also has on its upper face the mortises and the square mortise for the joint with the main mast step. Between frames M115 and M116 an opening of 28 × 22 cm was cut and fitted with a cover with bevelled edges, giving access to the bilge for cleaning in the section before the bilge pump. On its underside a series of longitudinal and triangular notches allow it to be firmly joined to the sister keelsons and to the frames over the keel. The two edges of the underside are cut lower than the longitudinal centre of the keelson. The keelson rests directly on the sister keelsons. From 1.50 m from the after end it has a series of triangular notches, two on each side, corresponding to symmetrical ones in the sister keelsons (Figure 4). At the level of the foremast cavity it has a series of transverse notches on its underside, as in wrecks of the Republican period, to fit directly not onto the sister keelsons but onto the upper face of the floor timbers.
The sister keelsons (Pinus halepensis) serve to join the keelson to the framing (Figure 4). They are 11.20 m long, with a constant section 11 cm high and 18 cm wide at M131; at the stern (M108) they are 13 cm high and 20 cm wide. Their forward ends are just at the position of the ship’s second mast. They have the corresponding notches for the keelson that fits over them. Between frames M124 and M125 a cross-tie joins them, which also helps to lock and secure the keelson. At their after end, their facing side faces describe a circular shape to create the space for the bilge pump, as in Port Vendres 1. The area is enclosed by a box of horizontal planks 5 cm thick nailed to substantial vertical posts of square or rectangular section (e.g. 26 × 24 cm). As usual in wrecks at accessible depths, the bilge pump was salvaged in antiquity (Figure 5).
Figure 3. Longitudinal section of the remains and reconstruction of the hull lines (after Liou and Gassend 1990: 223).
Figure 4. Mast step, keelson and sister keelsons (after Liou and Gassend 1990: 241).
Figure 5. Top view of the sister keelsons and the semicircles in their inner faces leaving room for the bilge pump. The four posts of the box that housed the bilge pump can be seen (after Liou and Gassend 1990: 254).
Frames
Fifty-five framing timbers survive. The moulded dimension of the frames is very constant (13 cm), and the sided dimension is between 14 and 16 cm. The room-and-space (malla), also very regular, is between 28 and 30 cm. The alternation of floor timbers and half-frames is rhythmic and is broken only at the bow, where three floor timbers of very closed lines follow one another (Figure 6).
In the widest part of the wreck, at frames M125–129, there is an alternation that has been described as asymmetrical (Liou and Gassend 1990: 219). Floor timber M127, defined as the master frame, has a total length of 3.24 m, 1.70 m to starboard and 1.52 m to port. M125 is 3.30 m long, with 1.70 m to starboard and 1.60 m to port, and M129 is 3.34 m long, with 1.70 m to starboard and 1.64 m to port. M126 and M128 also cross the keel, but the first is 3.50 m long, with 0.90 m to starboard and 2.60 m to port, and the second, with 3.30 m preserved, has 0.86 m to starboard and 2.44 m to port; so these cannot strictly be called offset half-frames, but rather asymmetrical floor timbers (Figures 7 to 9).
Futtocks occur throughout the hull. Notably, in the frames that cross the keel (floor timbers and asymmetrical floor timbers), the joints with the futtocks lie regularly along a line which in the central part of the hull is 1.60 m from the axis of the keel and gradually approaches it towards the ends of the hull. The joints between half-frames and futtocks lie instead along a line 1.50 m from the centreline. The position of these two longitudinal lines of joints is explained by the fact that the half-frames are practically as long as the floor timbers, which shows that the builders did not want to leave any part of the hull transversely weakened at the turn of the bilge. The joint has no scarf, but there are wedges.
The frames are fastened to the hull by treenails 1.5 cm in diameter, notably placed in groups of two,2 each pair 15–20 cm apart. At the bow, the last two surviving floor timbers were fastened to the hull by copper nails spaced 30 cm apart, one nail being at the level of the eighth strake. Three floor timbers were fastened to the keel by clenched copper bolts: M148 (Figure 3, broche) at the bow and M116 and M110 at the stern, although there were other copper bolts at the trait de Jupiter scarfs.
The limber holes are central, except in the floor timbers that coincide with the bolts, which have their limber holes at the sides of the heel. Finally, M142 was found to be a keel reused as a frame, as was M140, which has holes unrelated to its construction, apparently also a reuse.
Figure 6. Cross-sections of the hull from the bow of the ship, drawn looking aft (starboard on the left of the figure). From top to bottom: M155, M154, M153 (half-frame), M152 and M147 (after Liou and Gassend 1990: 224).
Figure 7. Cross-sections of the hull (starboard on the right of the figure). From top to bottom: M145, M140, M139, M135. The last section is drawn looking forward (after Liou and Gassend 1990: 225).
Figure 8. Cross-sections of the hull drawn looking forward. From top to bottom: M131, M127, M122; starboard on the right of the figure (after Liou and Gassend 1990: 226).
Figure 9. Cross-sections of the hull (starboard on the right of the figure). From top to bottom: M155, M118, M114, M108; the sections looking forward are M101 and M100 (after Liou and Gassend 1990: 227).
Planking
At the master frame M127, the hull (Pinus halepensis) preserves 18 planks on the starboard side, from the garboard strake to the lower wale. On the port side 4 strakes plus the garboard strake survive. The planks are 30 to 20 cm wide, narrowing gradually towards the ends of the hull, although among the stealers (tablas perdidas) at the stem the second starboard strake widens. The lengths of plank are between 6 and 9 m, some reaching 9 m, such as the second strake on the starboard side. The garboard strake, nearest the keel, is 7.5 cm thick, decreasing to 6.5 cm at the face in contact with the second plank, which has a constant thickness of 6 cm. The third ranges from 4.5 cm to 3.5 cm, which is the constant thickness of all the planks, except of course the wales, which are 9 cm thick and 16.5 cm wide (the 18th strake from the keel), corresponding to the waterline (Liou and Gassend 1990: 232). Two lengths of the same strake are joined by bevelled scarfs. The particularly thick garboard strake is joined to the keel by two rows of mortises and tenons placed in a sequenced rhythm (Figure 10).
The mortises in the strakes are spaced 12 cm apart and are 7 cm wide, slightly more than the tenons to allow some play when the planks are joined. They are 13 cm deep and 0.7 cm thick. The pegs are of truncated-cone shape, with a small diameter of 0.5–0.6 cm and a large diameter of 1.2 cm, and are driven from the inside of the hull, although three pegs driven from the outside were recorded at frames M147 and M155 joining planks 7 and 8, two at frame M148 with the metal bolt and one at M150. The frame physically fastened to the keel would have been joined to the hull before the hull was raised, thus possibly playing an active role in the ship’s lines3 (Figures 11 and 12). In two areas of the hull studied in detail, at the bow between M149 and M154 (Figure 13) and in the central area between M129 and M137 (Figure 14), small square-section pegs, 6 mm a side, were found, driven from the outside of the hull almost at the seam between two strakes and in the space between two frames. As in the Madrague de Giens, they were interpreted as pegs used to fix temporary frames to the hull to facilitate assembly.
Figure 10. Section of the garboard strakes, with the two rows of tenons placed rhythmically to join them to the keel. In the background, a floor timber with its limber hole in the centre (after Liou and Gassend 1990: 231).
Figure 11. Underside of a frame, with two pegs from the seam of the hull strakes embedded in it (after Liou and Gassend 1990: 239).
Figure 12. Upper face of the hull, showing how one of the pegs driven from the outside has been bent by the presence of the corresponding frame in situ (after Liou and Gassend 1990: 239).
Figure 13. Part of the hull at the bow, starboard side, raised to the surface in 1984 for detailed study (after Liou and Gassend 1990: 235).
Figure 14. Plan of the starboard side (after Liou and Gassend 1990: 235).
Ceiling, thick stuff, and wales
The ceiling consists of a fixed stringer nailed to the upper face of the frames, with lengths joined by bevelled scarfs. The stringers are on average 20 cm wide and 8 cm thick, 4.50 m long, placed at intervals between courses of removable ceiling strakes, which in the Saint Gervais 3 survived only as traces because of their poor state. The lower wale is 9 cm thick and 16.5 cm wide.
Caulking
The hull was coated internally with pitch.
Size and scantlings
The ship was reconstructed by Roman at 17 m in length and 80 t burden. The stanchion fitted at 60° in the mast step allowed the depth to be reconstructed at 2.60 m from the keelson, or 2.80 m in total; the master frame gives a maximum beam of 7.50 m for a length of 17 m, a ratio of 2.26.
Table 1. Scantling of the timber remains of the Saint Gervais 3 Shipwreck
| Timber | Sided [cm] | Molded [cm] |
| Keel | 14–17 | 18 |
| Sternpost | 15–18 | 22 |
| Keelson Sister keelsons | 47 (avg.) 18–20 | 27 11–13 |
| Maststep Mast mortise | 38 10 × 30 | 21 9 (depth) |
| Buttresses | – | – |
| Frames | 14–16 | 13 |
| Futtocks | – | – |
| Room-and-space At central frames | 28–30 | |
| Planking Garboard Second strake | 20–30 – – | 3.5 7.5 to 6.5 6 |
| Wale | 16.5 | 9 |
| Ceiling (stringers) | 20 (avg.) | 8 |
| Main mast Ø Chock | – – | – – |
Mast step 4.80 m long; keelson 10.50 m; sister keelsons 11.20 m. Planking, wale and ceiling: sided = width, moulded = thickness.
Wood
The species recorded are varied: Juglans sp., Acer campestre, Ulmus campestris, Pinus halepensis, Pinus pinaster and Pinus sylvestris. Keel, keelson, sister keelsons and planking: Aleppo pine (Pinus halepensis). Mast step: silver fir (Abies alba).
Bibliography
Liou, B. and Gassend, J.-M. 1990. L’épave Saint-Gervais 3 à Fos-sur-Mer (milieu du IIe siècle ap. J.-C.). Inscriptions peintes sur amphores de Bétique. Vestiges de la coque. Archaeonautica 10: 157–264.
Pomey, P. 1998a. Conception et réalisation des navires dans l’Antiquité méditerranéenne. In E. Rieth (ed.), Concevoir et construire les navires. De la trière au picoteux. Technologies, Idéologies, Pratiques 13.1: 49–72. Ramonville Saint-Agne: Érès.
Pomey, P. and Rieth, E. 2005. L’archéologie navale. Paris: Errance.
Notes
1 Its 60° inclination allowed the depth of the hull to be reconstructed at 2.60 m from the keelson, giving a total of 2.80 m. The master frame gives a maximum beam of 7.50 m for a length of 17 m, a ratio of 2.26.
2 At the time of excavation this was wrongly associated with the Cavalière wreck.
3 It should be remembered that there is no change in the longitudinal design principle of the ship, only an aid in the building process. It could be due to the reproduction of a known model, using templates to give it shape.