Madrague de Giens, 75–60 BC
Country: France
Place: 350 m off the Pointe de l’Ermitage, Giens peninsula (Var), at a depth of 18–20 m
Coordinates (approximate): 43.04, 6.10
Type: Large Roman Republican merchant ship with double planking
Identified: No
Dated: 75–60 BC, based on the study of the cargo.
Beam: 9 m (Pomey 1982: 146).
Keel Length: Unknown.
Length Overall: 40 m (35.10 m preserved) (Pomey 1982: 146).
Number of Masts: At least two (main mast and foremast); a third, aft, is considered likely.
History of the wreck
The Madrague de Giens wreck was discovered in 1967. Its systematic excavation between 1972 and 1982 by Pomey and Tchernia is a landmark of underwater archaeology, not only in France but internationally, during which excavation methods and techniques still in use today were developed. The site was reopened in 1993 to take samples for dendrochronological analysis. The first four campaigns, from 1972 to 1975, were carried out in the central part of the wreck; from 1976 the excavations covered both bow and stern, with important results that justified further campaigns until 1982 (Pomey 1982: 133).
During the excavation a number of stones of local origin were found associated with holes dug into the cargo and in the area of the bilge pump, which would indicate that in antiquity urinatores1 salvaged part of the cargo and of the ship’s equipment.
The Madrague de Giens shows how medium-sized and large ships were built to perform with complete efficiency at sea, with a structure able to withstand the stresses of navigation. This may seem self-evident, but the Madrague de Giens confirms it, thanks to the customary knowledge of the master shipwrights. In great works, nothing was left to improvisation.
Description of the site
The wreck lies at a depth of 18–20 m and only 350 m from the Pointe de l’Ermitage, on the Giens peninsula. The site was intact at the time of its discovery. The stern area had lost the upper parts of the hull, despite the 2–3 m layer of sediment covering it.
Cargo
The cargo consisted of between 6,000 and 6,500 Dressel 1B wine amphorae of four different morphological types, stowed in a staggered pattern in three superimposed layers. Many of them bore the stamps of the production workshop, the most frequent being that of Publius Veveius Papus, which allowed the region of origin to be located at Terracina (Hesnard 1977), south of Rome. The amphorae are thought to have carried Caecuban, one of the finest Italian wines of the period. Campanian and common wares, packed in crates under the amphorae, travelled with the cargo. Volcanic sand from Campania and brushwood dunnage helped the stowage.
In the stern of the ship, tableware belonging to the crew was found: pieces of common ware and Campanian A ware2 with graffiti, some lamps, a simpulum, a bronze balance, three lead ingots, wheat and two helmets.
Ballast
Not reported. Volcanic sand from Campania was used in the stowage of the cargo.
Anchors
Not reported.
Guns
Not applicable.
Iron concretions
Not reported. Bronze bolts and copper nails were used in the construction.
Hull Remains
The hull is preserved over 35.10 m of its length. It has a double hull (two layers of planking) sheathed with lead, a keel with two rabbets, a long mast step continuing as a keelson, and a ceiling with fixed stringers and removable strakes (Figure 1). No wales were found, since the hull did not survive to that height; a tunnel dug from side to side of the hull confirmed their absence.
The study of the stern made it possible to understand some construction questions relating to the central part of the hull. At least two strakes of the main hull (the fourth and fifth) were fitted during construction, before the second hull or planking, and were later completely replaced (along their whole length), systematically on each side of the keel. This modification resulted in particular in the reversal of the direction in which the tenon-locking pegs were driven, which were driven from the outside of the hull. This reversal affects only these replaced strakes and the adjacent ones, and is independent of the general process of construction of the hull, in which the pegs are driven from the inside. It therefore has no relation to the presence of active frames in the construction of the Madrague de Giens.
Figure 1. Axonometric view of the central part of the hull of the Madrague de Giens (after Pomey and Rieth 2005: 109).
Keel, keelson and mast step
For the study of the keel, a 1 m section was raised from a central part of the site, coinciding with the maximum beam (Figure 2). Its section drawing shows that it was 39–40 cm high, with a maximum width of 35 cm at the level of the second rabbet. The first, notched rabbet, intended to receive the inner hull, is cut practically at right angles from the upper face of the keel and is 10 cm high and 9 cm wide; it leaves the upper face of the keel 16 cm wide, right in the bilge. The second rabbet is higher, 14 cm, but much narrower, only 4 cm.
In more than ten cases there are bronze bolts, 3 cm in diameter, joining the central part of the floor timbers to the keel, driven from the outside and closed by a square rove 5 to 6 cm a side. All the floor timbers from 100 to 116 have notches 55 to 57 cm long, 2 to 3 cm wide and 5 to 7 cm high, placed alternately on each face from one floor timber to the next, facing each other. Into them is fitted the mast step, 55 cm wide, 45 cm high and nearly 4 m long, which has on its underside perpendicular grooves of the same width as the floor timbers, regularly spaced to fit over them. Its lower corners are bevelled in a circular arc to make the pieces easier to fit. It is held by its own weight, and the notches prevent lateral or longitudinal movement of this piece (Figure 3). Behind the mast step proper there continues a keelson, 19 cm wide and 15.5 cm high, with notches for the floor timbers in its underside. Unlike for the mast step proper, the floor timbers have no notches to receive the keelson.
In the stern, the bottom of the hull had flattened lines towards the centreline, rising gradually. At 10.50 m from the end of the preserved keel, it rises at a constant 16.5° above the horizontal before beginning a much sharper curve at the start of the sternpost, creating a large angle which indicates that the sternpost assembly far exceeded the 10.5 m preserved, certainly in the order of 14 to 15 m, larger than in any other known wreck (Figure 4).
The rake of the hull lines at the stern meant that the backbone had to withstand very great mechanical stresses (Pomey 1982: 136), so it needed corresponding strength and rigidity, which posed enormous construction problems. The detailed study of the backbone at the stern, thanks above all to the sections raised to the surface, has made it possible to understand the ingenious design ideas of the builders. One of the particularities of the enormous size of the Madrague de Giens ship is that the natural limits on the dimensions of timbers obliged the builders, as in the modern and contemporary periods, to make structural parts of the ship from successive pieces, since they could not be made from a single timber (Figure 5).
Five pieces form the sternpost assembly. The keel (Figure 5, 1), which begins to rise, is joined to a connecting section with the sternpost (Figure 5, 3) and is reinforced above by a deadwood block (Figure 5, 2), creating a composite “keel” of several large pieces. The rabbet3 of this piece, composed by adding the deadwood block above, is notched, with an angle that somewhat flattens the rise of the garboard strakes. A second, straight connecting section of the keel is joined by a very complex trait de Jupiter (rayo de Júpiter) scarf, secured with a horizontal key, to the sternpost proper, with an inner sternpost fitted on the inside just where the change of angle for the rise of the timbers is greatest (Figure 5). Likewise, a false sternpost was fitted on the outside, covering the whole length of the joints between the sections of keel and sternpost (Figure 6). These timbers not only withstood the mechanical stresses on the after part of the hull, but also gave the ship important sailing advantages by forming a large skeg. It is located in the rising section of the sternpost (quête), where the hull lines are practically flat, unlike the central part of the hull, where they are strongly pinched and leave the keel in a very prominent position. The combination of the shape of the sternpost and of the keel itself in the central part of the hull gave the ship great stability in all conditions, by increasing the total lateral plane. A skeg was also recorded in wrecks such as the Marsala and Dramont A.
The excavations at the bow of the ship, begun in 1980, deliberately left a 4 m section as a control, since this part of the hull was already well known, unlike the bow, which was of great interest. The working hypothesis was that the Madrague de Giens might be one of the ships with a concave bow known from iconography, such as the mosaic of the Syllectini in the Piazzale delle Corporazioni at Ostia, which shows a sternpost forming a characteristic inclined plane and a particular bow. An area 8.80 m long was excavated, in which the preserved timbers showed how the hull lines closed abruptly, creating an almost “pointed” bow. The excavation documented the forward section of the mast step, already recorded aft and in the central part of the hull (Figure 7). With a total length of 8.10 m, the mast step is notched over the floor timbers and, by means of mortises cut for the purpose, supports the feet of the masts while distributing the mechanical stresses they cause throughout the hull. Besides the central cavity, it had another forward for a second mast. However, the arrangement of the cavities or mortises and the characteristics of the cargo made it impossible to step an inclined mast as seen in iconography, so it seems more likely that a vertical stanchion was placed in this mortise, against which the foremast would have rested at deck level with the necessary inclination (Figure 11).
As at the stern, the bow rises progressively from a point 7 m from the beginning of the stem, just ahead of the start of the mast step, and despite the rake of the stem the two ends of the hull are not symmetrical. The stem proper preserves only 40 cm of height (since it is concave), being joined directly to a connecting piece (forefoot) with the keel. It is raked at 62° to the horizontal, facing aft. To protect and strengthen this piece, a cutwater reinforces it at the front. Only its lower part survives, 1.45 m long, but it can be reconstructed from the shape of the stem, forming a pronounced concave line (Figure 7). The “reversed” stem with the cutwater would not exceed 9.50 m in total length. The angle of rise of the keel at the bow is gentler (10°) than at the stern (16.5°). The Madrague de Giens is one of the few archaeological examples of ships with a concave bow, which is attested in iconography less often than convex bows or “symmetrical hulls”.
The shape of its cutwater may seem surprising in a merchant ship, but a military function is entirely ruled out, both by the characteristics and function of the ship itself (a large sailing ship) and by the limited strength of this bow structure, in no way intended to receive heavy impacts, with a simple pegged tenon in the stem and mortises cut in the cutwater. Its explanation, as with the stern, is related to better sailing qualities and to the strength of the backbone, and also to the use of cargo space (there was cargo right into the bow), since the closing of the lines towards the bow can be delayed as long as possible. The sharp closing of the bow and the resulting “artificial” increase in length also increases the lateral plane. It also acts as an element that modifies the formation of the bow wave, reducing resistance to forward motion and thus improving speed.
Figure 2. Cross-section of the keel with the first strakes of the two hulls and the floor timber (after Tchernia et al. 1978: 77).
Figure 3. Cross-section at the mast step, showing the system of fastening the timbers together (after Pomey and Rieth 2005: 128).
Figure 4. Longitudinal section of the stern of the Madrague de Giens. On the right the mast step, in the centre the casing of the bilge pump, and on the left the angle of rise of the sternpost (after Pomey 1982: 135).
Figure 5. Exploded view of the sternpost of the Madrague de Giens. 1: keel with simple bevelled rabbet; 2: deadwood block with polygonal underside which, joined to the keel by tenons, creates a kind of notched rabbet for the rise of the garboard strakes of a flat, strongly raked stern; 3: connecting piece between keel and sternpost, whose rabbet makes the transition between that formed by the keel and deadwood and the triangular notched rabbet of the sternpost; 4: sternpost, joined to the connecting piece by a trait de Jupiter scarf; 5: inner sternpost reinforcing the curve of this piece; 6: false sternpost forming the ship’s skeg (by the author, after Pomey 1982: 137).
Figure 6. Photograph of the transverse section through the proximal part of the sternpost assembly. 1: keel with simple bevelled rabbet; 2: deadwood block; 6: false sternpost forming the ship’s skeg (by the author, after a photograph at www.ccj.univ-provence.fr).
Figure 7. Longitudinal section of the bow. 1: mast step; 2: bolt fastening the framing to the keel; 3: keel; 4: concave stem; 5: cutwater; 6: reconstructed profile of the bow (by the author, after Pomey 1982: 143).
Frames
The frames vary in sided dimension from 9 to 18 cm, 13–14 cm being the most frequent. The room-and-space (malla) is 10 to 11 cm on average, with a regular alternation of floor timbers and half-frames. The floor timbers, between 3 and 4 m long, have a very triangular profile in their central part, 57 to 60 cm moulded, decreasing to 12 to 15 cm one metre from the centreline (Figure 3). They are not symmetrical. To let water circulate in the bilge, the heel is not in direct contact with the keel, and some have a central triangular limber hole; others have a circular hole drilled with an auger (5 cm in diameter). The floor timbers are not in direct contact with the futtocks but are separated from them by 2 cm, with a horizontal or bevelled cut.
The framing is fastened to the hull by truncated-cone wooden treenails, 2 to 2.5 cm in diameter, driven from the outside of the hull, two treenails per strake and frame. At the level of the first three strakes, the floor timbers are not fastened by treenails but by nails driven from the outside of the first hull. Nailing later joined the two hulls. There is a clear rhythm in the placing of the treenails and the blind nails fastening the frames, so that they do not coincide and weaken the piece.
Planking
The garboard strake of the inner hull has a pentagonal section, 12 cm thick. The second strake has a rectangular, slightly curved profile, 10 cm thick against the garboard strake, decreasing to 8 cm at its face in contact with the third strake; from the fourth strake the thickness is a constant 6 cm. The outer hull is a constant 4 cm thick, giving a total thickness of 10 cm. The inner strakes are 18 to 32.5 cm wide, the most common widths being 20 to 23 cm. These measurements could not be taken systematically on the outer hull because of the lead sheathing, but where observation was possible the values were similar to those of the inner planking.
In the inner hull (Figure 8) the mortises were 8 to 8.5 cm wide, 10 to 12 cm deep and 1.2 to 1.5 cm thick, regularly spaced every 6.5–7.5 cm. Those of the outer hull are 5.5 to 5.7 cm wide, 6 cm deep and 0.7 to 0.8 cm thick, with a greater spacing of 9.5 to 10 cm. The locking pegs are 1.5 to 1.8 cm in diameter in the inner hull and 0.5 to 1 cm in the outer hull. The tenons are generally slightly smaller than the mortises in which they are housed. In the keel the mortises are logically very large: 5.5 to 6.6 cm wide, 6 to 7 cm deep and 0.8 cm thick for the outer hull, and 14 cm deep and 1.6 cm thick for the inner hull. The staggered arrangement is not always respected. Two lengths of plank in the same strake are joined by bevelled scarfs, with horizontal nails improving the joint.
Figure 8. Plan of the joints between the planking strakes in the inner hull (I) and the outer hull (II, III), showing the number and size of the tenons according to the part of the hull. Horizontal reinforcing nails can be seen in the bevelled scarf (after Tchernia et al. 1978: 79).
Ceiling, thick stuff, and wales
The ceiling of the hold, with a robust stringer, is the only longitudinal reinforcement preserved on the wreck. The stringers were fastened to the frames with copper nails, without any notches in their underside. They have a rectangular or slightly trapezoidal section, 20 to 30 cm wide and 6 to 10 cm thick. The joints between two lengths of the same piece are bevelled, about 5 cm long. Between the fixed stringers there are courses of removable ceiling strakes simply laid on the frames, narrower (15 to 25 cm), thinner (2.4 to 4 cm) and 2 to 3 m long.
The first stringer on each side of the mast step is somewhat different from the others. It is 15 cm wide, and a removable ceiling strake of the same thickness (8 cm), 10 cm wide, is inserted between the stringer and the mast step. This arrangement would have allowed access to the bottom of the bilge despite the mast step. There is also a ceiling on the centreline of the hull, made of removable planks perpendicular to the keel, 70 cm long and 6 cm thick, of somewhat irregular width (6 to 29 cm), resting on the first fixed stringers (Figure 1). Five stanchions supporting the deck were recorded, 16 to 18.5 cm in diameter, with their bases roughly squared with an adze.
In the stern area, without amphora cargo, where the ship’s living quarters were presumed to be, remains of timbers and box-shaped planking related to the bilge pump were found, located just before the beginning of the rise of the backbone towards the stern (Figure 4). The pump was salvaged in antiquity.
Caulking
The lead sheathing covering the hull (Figure 9), as in other recorded cases (e.g. Colonia Sant Jordi), consists of sheets 2 mm thick, nailed in a staggered pattern every 5.5 cm with small nails with large flat heads, over a layer of tarred woven wool (with other known examples, e.g. Nemi). The presence of Posidonia oceanica had turned the lead almost entirely into sulphide. The whole inner and outer hull was coated with hot resin, as was the outer face in contact with the lead sheets. The resin accumulated in the bilge, even blocking limber holes.
Figure 9. Axonometric view of the bottom of the hull, showing the keel with its two rabbets for the inner and outer hulls, the joints between the two impregnated with resin, and finally a layer of wool with resin on the outside of the hull, which in turn receives the nailed lead sheathing (after Pomey and Rieth 2005: 121).
Size and scantlings
The excavations have allowed the proportions of the ship to be reconstructed. With 35.10 m preserved, its overall length is 40 m, its maximum beam 9 m and its depth in hold of the order of 4.50 m, giving a burden of 375 t (Pomey 1982: 146). The length-to-beam ratio is more than 4 (length/beam = 4.4), higher than that of other ships, admittedly of more modest size, whose coefficients are closer to 3. The higher the coefficient, the greater the potential speed of the sailing ship. The strongly raked stern and the low bow would have improved the ship’s behaviour, while the position of the centre of lateral resistance, with the main mast at 5/8 of the length and the foremast at ¼, implied a very forward position of the centre of effort of the sails, in keeping with the centre of lateral resistance of the hull (Pomey 1982: 148). This sail arrangement, together with the shape of the hull, must have given the ship greater stability and speed. The hypothesis of a third mast at the stern, by analogy with iconography, is more than plausible, also in terms of the trim and balance of the hull. These square sails and the large lateral plane surely allowed the Madrague de Giens to sail to windward like square-rigged sailing ships of the 19th century (Pomey 1982: 152).
The similarity between the construction of the Madrague de Giens and the ship in the mosaic of Thémétra (Tunisia), of the mid-3rd century AD (Figure 10), is astonishing, even the proportions of the hull coinciding. This shows that iconographic representations are sometimes very accurate and faithful, so the comparison between the Madrague de Giens and this iconography can be taken further, since it is not a fanciful representation. The rig, above all as regards the foremast, differs greatly from that shown in other iconography, especially in its size, which would have given first-rate propulsion besides its role in balancing the ship. The Ostia and Thémétra representations date respectively from the 2nd and 3rd centuries AD, so large sailing ships such as the Madrague de Giens should perhaps not be considered exclusive to the Hellenistic period, although for the time being there is no archaeological evidence of large ships of asymmetrical form in the Imperial period (the Saint Gervais 3 cannot be compared because of its short length and small burden). Equally, in the 1st century BC this was a highly evolved type of ship, developed from models that could go back decades.
Figure 10. Example of a large ship whose length-to-beam ratio produces an elongated merchant ship. Mosaic of Thémétra, Tunisia (after Pomey et al. 1997: 89).
Figure 11. Comparison of the proportions of the ship in the Thémétra mosaic and the Madrague de Giens wreck. The position of the main mast, together with the angles of bow and stern, reveal how the model of ship matches the proportions of the wreck (after Pomey 1982: 147).
Table 1. Scantling of the timber remains of the Madrague de Giens Shipwreck
| Timber | Sided [cm] | Molded [cm] |
| Keel | 35 (max.) | 39–40 |
| Keelson (aft of mast step) | 19 | 15.5 |
| Maststep | 55 | 45 |
| Buttresses | – | – |
| Floor timbers | 9–18 (mostly 13–14) | 57–60 (centre) 12–15 (1 m from centreline) |
| Futtocks | – | – |
| Room-and-space At central frames | 10–11 (avg.) | |
| Planking (inner) Inner garboard Outer planking | 18–32.5 – – | 6 12 4 |
| Ceiling Stringers Removable strakes | 20–30 15–25 | 6–10 2.4–4 |
| Deck stanchions (Ø) | 16–18.5 | – |
| Main mast Ø Chock | – – | – – |
The mast step is 8.10 m long in total. Planking and ceiling: sided = width, moulded = thickness; total hull thickness (inner + outer planking) 10 cm.
Wood
Not reported.
Bibliography
Hesnard, A. 1977. Note sur un atelier d’amphores Dr. 1 et Dr. 2-4 près de Terracine. Mélanges de l’École française de Rome. Antiquité 89.1: 157–168.
Pomey, P. 1982. Le navire romain de la Madrague de Giens. Comptes rendus des séances de l’Académie des Inscriptions et Belles-Lettres 126.1: 133–154.
Pomey, P. (ed.), Gianfrotta, P. A., Nieto, X. and Tchernia, A. 1997. La navigation dans l’Antiquité. Aix-en-Provence: Édisud.
Pomey, P. and Rieth, E. 2005. L’archéologie navale. Paris: Errance.
Tchernia, A., Pomey, P., Hesnard, A. et al. 1978. L’épave romaine de la Madrague de Giens (Var) (campagnes 1972–1975). Gallia suppl. 34. Paris: CNRS.
Notes
1 The stele from the port of Ostia dedicated to the emperor Antoninus Pius, of the 2nd century AD, reveals the existence of a guild of specialised divers called urinatores.
2 The Campanian ware consisted mainly of light-clay forms, Lamboglia 1, 2, 3, 4, 5/7 and 11. In the area of the bilge pump, large plates of grey-fabric Lamboglia form 5/7 were found, some still stacked, together with fragments of the boards that formed the packing crates for this pottery.
3 The rabbets are simple bevels in the keel proper, but the keel can in fact be considered to have notched rabbets if the internal deadwood block is added.