1Since the discovery of the famous Ljubljana (Lipe) cargo ship of sewn construction (1890), other important remains of large capacity watercraft have come to light in the area of the Ljubljansko barje (Ljubljana Marshes, central Slovenia), particularly in the Ljubljanica River at ancient Nauportus (present-day Vrhnika) (fig. 1). These offer physical evidence of the writings of the ancient Greek geographer Strabo (Geography 4, 6, 1; 7, 5, 2) who records large-scale waterborne cargo traffic between the Latin colony of Aquileia, on the one side, and the Transalpine Celts in the hinterland of the northern Adriatic and the peoples of the Danubian provinces on the other (Šašel Kos 1990, p. 17-21, 143-148).

2This paper focuses on the results of an underwater salvage excavation in the Ljubljanica riverbed at Vrhnika, conducted in 2015. These revealed a highly interesting cluster of finds from the late 2nd to the first half of the 1st century BC (Gaspari et al. 2017). The cluster is most probably associated with the presumed cargo and reloading post of pre-Roman Nauportus. It consists of a large capacity oak logboat, the wreck of a roughly contemporary sewn cargo ship and the skeletal remains of an adult woman from the early 1st century BC lodged in the sediments directly under the logboat (fig. 2).

3Geoarchaeological investigations at the site have indicated that in comparison to the present the riverbed was broader in this section during the Iron Age. They have also shown that the archaeological context was formed in the sandy sediments of the gently inclined underwater slope along the right bank. The logboat, which the lateral shift of the channel caused to break into three parts, was obviously abandoned here. It was intentionally immersed complete, in a functional state. As for the sewn boat, the initial investigation provided no solid evidence as to why it ended up on the bottom of the river. Further objects related to shipbuilding activities lay exposed on the river bottom, including a board with an unlocked mortise-and- tenon joint (radiocarbon dated to the 3rd-2nd century BC) and a large number of iron clamps. A significant part of the latter might be understood as the remnants of post-Roman shipping, since the harbour of medieval and early post-medieval Vrhnika (Oberlaibach in German) occupied more or less the same area as the presumed pre-Roman reloading and toll station, with piers and associated facilities, platforms and shipyards.

4At the end of the investigations, the logboat was lifted and transported to conservation facilities, and the detached wooden elements were collected, while the sewn boat remains were left in situ.

1. Boats from the Ljubljanica River in the Area of Pre- and Early Roman Nauportus

1.1. Logboat

5The logboat is one of the largest known late prehistoric and Roman examples of this type of watercraft. It originally measured approximately 14.4 m in length, 1.25 m in beam and 0.68 m in depth. It was made of a roughly 200 year-old oak tree (Quercus sp.), with the lower part of the trunk used for the bow (Gaspari et al. 2017). The characteristic sections have rounded sides and a flat bottom, on both the inner and outer surfaces. The more probable of the two possible displacement situations calculated from the reconstructed lines and sections of the boat with a light weight at around 1.5 tonnes in a partly saturated condition and with a cargo space volume of around 6 m3 shows a total load capacity of around 2 tonnes at a still safe draught, retaining 20 cm of freeboard. Considering the asymmetric shape with the centre of gravity in the front part and a protrusive stern, the logboat was probably powered by paddle and/or punt, and, if necessary, by towing from the bank.

6In addition to its large size and capacity, the logboat also stands out in its technological features. There are control perforations, but even more particular is the substantial use of iron clamps and fittings, wooden treenails, plugs, patches and boards, as well as resin-soaked woollen cloth for repairing cracks and extensive luting of the replaced parts.

7The bottom of the boat has two openings along the mid-axis. The one towards the stern is perfectly round, with a diameter of 12 cm and oblique walls accommodating a conical plug secured with an iron clamp. The other opening, located between the first and second thirds of the boat’s length (fig. 3), is of a more complex design. A large oval patch board with truncated ends and a stepped section closes the opening and was lined with cloth previously soaked in resin. The cloth was held in place with two cylindrical treenails reaching through the main trunk and secured with three iron clamps from the inside and two more from the outside. The function of the openings is debatable. The larger one, without any signs of subsequent removal of the board, probably replaced a weakened part of the bottom at a knot, while the round one at the rear may have served for discharging bilge water and/or facilitating immersion for storage, though it may also have been conceived as a lock, preventing the theft of the craft (see Medas 1994, p. 35).

8The replacement of the bow section is particularly interesting: the original was probably weakened due to the growth pattern of the root part of the trunk. It was replaced with a perfectly fitting block cut from the same trunk, indicating that the intervention took place in the initial phase of the logboat construction. Similar to the design of the above-mentioned board, the surface of the main trunk was caulked with large resin-soaked cloths before installing the substitute part that was subsequently attached with cylindrical treenails fitted in the pre-prepared recesses in the corresponding parts of the bow. The final stage of repair consisted of fastening the protruding extension of the prow by adding a band-shaped clamp and securing the cracked areas of the main trunk with nailed fittings and additional clamps. The textile used for waterproofing was made in plain weave technique (tabby y/y) from Z-twisted wool yarn, corresponding to the technique of the tablet woven textiles from the Middle and Late La Tène sites in central Europe, as well as textiles from pre-Roman and Roman contexts in the eastern Po Plain.

9A singular feature is a forged iron nail with a hemispherical head that resembles the rivets of the roughly contemporary (second half of the 2nd-early 1st century BC) Celtic shield bosses, both those with a rectangular flange and a barrel-shaped dome and those of the Mokronog-Arqua type, with a main distribution area between the central Transpadana and southeastern Alpine regions.

10Luting with cloth soaked in a natural resin is typical of ancient Mediterranean shipbuilding, which supposedly also influenced such practice in the Roman-era vessels of the Rhône basin (Médard 2011). While joining with coaks/treenails is a very old technological solution, known at least as early as the Late Stone Age in continental and Atlantic Europe, the extensive use of iron clamps may indicate a local, Celtic tradition. It became more widespread later on with the use of numerous small iron clamps in the shipbuilding process of the Early Roman Period. Both large holes along the logboat’s central axis are similar to those in two presumably Roman logboats from the Po Delta (Berti 1987, p. 20, 23, figs 6 and 12). Together with other details, this would indicate an exchange of technological knowledge within the northern Adriatic region.

11Found next to the logboat towards the middle of the riverbed was an oak board with the remains of an iron clamp, two repairs of cracks with resin-soaked textile and the remains of an unpegged mortise-and-tenon joint. Radiometric measurements have shown it to be earlier than the logboat (fig. 2), but the board may nevertheless represent the repair of one of the sides at the stern: we may verify this when physically comparing the pieces after the completion of the logboat’s conservation.

1.2. Sewn Boat

12The upper part of a wreck’s side was discovered just north of the logboat’s bow at the foot of the riverbank. The research of the wreck was limited to preliminary documentation of the exposed top section, situated diagonally to the riverbed (Gaspari et al. 2017, p. 85-87). A 2.15 m-long part of the side planking and frames was recorded (fig. 3). The boat end protruding from the bank was destroyed through post-depositional processes and the condition of the part embedded in the bank is unknown. The topmost of the surviving planks is 3 cm thick: its original width is unknown. It is presumably made of conifer wood and bears obliquely drilled sewing holes in intervals of 7-8 cm, approaching the prevalent values of 7.5-8.5 (some reaching up to 10) cm recorded on the Ljubljana (Lipe) boat (see below). The direction of the holes corresponds with the logic of minimised rope exposure on the external surface of the planking. Ropes for sewing and rolls of wadding placed on the plank joints were made of undetermined plant material, while the small pegs for blocking the rope and its reinforcement were made of yew wood (Taxus baccata). The surviving upper part of the central rib (of the three exposed) bears a semicircular notch (crenulation) in the axis of the wadding at plank joints. The ribs and cylindrical treenails for attaching the frame elements to the planking were made of oak (Quercus sp.). The upper parts of ribs measured 8 x 8 cm in section and were placed 0.6 m apart, measured from centre to centre, again resembling the framing interval values on the Ljubljana (Lipe) boat. The dimensions of particular elements and the observed constructional features indicate that the newly discovered remains belong to a large barge. It has been radiocarbon dated to the 2nd century BC (fig. 2).

2. The Ljubljana (Lipe) Sewn Boat: Surviving Evidence, Radiocarbon Age and Remarks on the Lesser-Known Technological Signatures

13Upon its accidental discovery in October 1890, the barge from Lipe was almost completely preserved. The excavation and documentation of the approximately 30 m long, 4.8 m wide and 0.6 m high boat followed the next spring. It was conducted by Alfons Müllner, curator at the Provincial Museum of Carniola in Ljubljana (Laibach in German) and his team. They removed two sections to preserve at least a partial insight into the vessel’s construction (Müllner 1892). As the successor institution, the National Museum of Slovenia presently holds 110 wooden elements, 20 iron nails (all reproduced in drawings in Karinja 1998), field drawings (technical ground plan with one general section and several sections of details, a more ‘artistic’ plan of a less informative value), two photographs (one taken shortly after discovery and the other before the conclusion of the fieldwork in April 1891) and a model of the ship in the scale of 1:10 (fig. 5). The model was skilfully executed by Karol Binder, museum carpenter, based on his own measurements of the remains in situ (Gaspari 1998a, 1998b). Müllner also provided a detailed description, which includes expert identification of the wood and plant species used in the construction. His study far exceeds the average quality of publications in his day, but does leave some constructional issues unanswered. One of the most intriguing questions, the date of the boat, long remained unclear despite a series of radiometric measurements. Recently, wiggle-matching calibration of two radiocarbon dates of a spruce (Picea sp.) plank with one of the longest surviving tree ring sequences has pointed to the ship’s construction between the mid-2nd and the mid-1st century BC (fig. 2). It thus confirms the initial hypothesis of Müllner, who proposed an Iron Age date based on stratigraphic evidence.

14The unique shape of the boat, clearly adapted to the specific environment of the Ljubljana Marshes, is characterised by an ellipsoidal-oval outline with truncated ends, a flat bottom and slanting sides. Two long strakes of boards (stringers) of ash wood (Fraxinus sp.), laid parallel to the ship mid-axis at a distance to one another and interpreted as members of a roll-on roll-off corridor, support the alleged use of the pramm for the transport of goods in barrels, such as is described in contemporary texts (Strabo, Geography 1, 8, c. 214).

15The technological characteristics of the Lipe (Ljubljana) ship have been discussed in a series of papers following Gerhard Salemke’s ‘rediscovery’ of the original publication in 1972. The present contribution therefore only highlights the observations and remarks on the lesser-known details based on the surviving parts, original drawings and the ship model. The first concerns the arrangement of the floor planking in longitudinal strakes with symmetrically distributed diagonal scarfs. Their pattern only survives on the model (fig. 5), but there is no reason to doubt the authenticity of the reproduction in spite of Müllner’s silence on the subject.

16There are uncertainties about the sewing technique employed in the boat as Müllner’s drawing (see the interpretation schemes proposed by Boetto and Rousse (2011, p. 184, fig. 7) presents it as of a continuous type in a zigzag pattern, while the model shows a diagonal (///) pattern. It is worth mentioning the absence of notches of any kind at the holes of the obliquely bored cylindrical channels on both surfaces. The alignment of the channels also requires further study, but a hint is provided by Müllner himself, who wrote that the planks were installed facing upwards, meaning with concave side of the tree rings facing the internal surface. This is of importance for understanding the direction of the channels as drawn by the excavator, which peculiarly expand towards the external surface of the ship’s bottom (Müllner 1892). This would result in extensive exposure of the rope to wear and tear during navigation. The observed growth pattern of the surviving planks is not always in accordance with the statement above, as both situations occur: sometimes the holes of the oblique channels do lie close or at the very edge (corner) of the alleged internal surface of the plank (with concave facing growth pattern), but also vice versa. The alignment of the channels as proposed by the excavator would be unique and odd. We should therefore consider the possibility that the detail was incorrectly drawn, as Boetto and Rousse (2011, p. 183-184) have already suggested.

17Bundles or rolls of lime (Tilia sp.) bast were placed above the plank joints. After the rope had been tightened, the channels were locked with pegs of common buckthorn (Rhamnus carthartica).

18There are several other features showing that the whole set of technological solutions inherent to the sewn boats of the Mediterranean tradition was adopted. The cylindrical recesses in the sides of three surviving planks (Nos. 6, 45, 100) testify to the use of dowels at intervals between 15 and 23 cm (Gaspari 1998a, 1998b; Karinja 1998, pls. 20.1 and 21.5). These were not extensively used, as the detail occurs neither in the description and drawing nor on other surviving elements. It is possible that they served to secure the (side?) planking, but their role in providing the longitudinal structural rigidity of the hull also cannot be excluded.

19The other common and frequently discussed feature of the Ljubljana boat is crenulated timbers (Quercus sp., Fraxinus sp.) and ribs (Quercus sp.). These enabled the frame elements to pass over the wadding placed along the seams of the longitudinal strakes. The frames, which occur at regular intervals of 60 cm (possibly adhering to the Italic/Roman measuring standards based on one Roman pes), were only fastened to the planking with cylindrical treenails (Rhamnus carthartica), although several large and small iron nails were also collected from the boat.

3. Discussion

20Of the newly acquired evidence of organised water transport on the Ljubljanica at the end of prehistory, the cultural and technological context of the discovered sewn ships is particularly interesting. Radiocarbon dates show the barges from Lipe and Vrhnika to be the earliest known examples of flat-bottomed sewn boats of the northern Adriatic area (the regional traditions recently analysed by Pomey and Boetto 2019), specially designed to navigate shallow, slow-moving or stagnant waters. This raises a number of questions. These concern the origin of the technology, its carriers or modalities of the transfer of artisanal solutions. They also pertain to the owners or users of such high-capacity vessels in the late prehistoric environment deep in the hinterland of Caput Adriae, in the period before the presence of the Italic/Roman merchants and the military became more intense around the mid-1st century BC.

21The roughly contemporaneous use of the sewing technology and mortise-and-tenon joints in the construction of local vessels already in the 2nd century BC, as well as resin-soaked textile luting of logboats, all seem to be rooted in Mediterranean shipbuilding tradition, which is of no surprise as the Ljubljansko barje is only located some 55 km from the nearest coast. As attested in literary sources for the final centuries BC, the sewing technique was employed by the neighbouring tribes of the Histri and Liburni, while the records on the river and lagoon of flat-bottomed vessels of the Veneti make no mention of their modes of assembly, although they might also involve ligatures. In connection with this, we should bear in mind the cultural and possibly even ethnic background of the local population at Nauportus. As argued recently (Gaspari et al. 2017, p. 179-185), the linguistic origin of autochthonous (pre-Roman) names indicates an integration of the broader area of pre-Roman Nauportus into the northern Adriatic onomastic region, consolidated around the Venetic, Histrian or Liburnian name tradition. The alleged kinship is corroborated by the local toponym Emona (present-day Ljubljana) with its typical northern Adriatic suffix -ona, as well as the similarities of the epigraphically attested local female deity Aecorna with the Venetic and northern Adriatic deities. The potential connection between the prevalence of specific artisanal, in this case shipbuilding knowledge, and the ethnic/linguistic affiliation of local community(ies) to a larger group of peoples in the Caput Adriae area is further supported by a strong non- Celtic component in the local Late Iron Age material culture. These indicate a close attachment to the (local) Late Hallstatt tradition and the integration of the coastal and hinterland areas between Friuli, Notranjska and the Kvarner Gulf or northern Dalmatia into the regional trends.

22In the cultural regionalisation of the La Tène Period, the area between Ljubljana and Vrhnika is traditionally considered a territory under the military-political authority of the Tauriscan tribal alliance. Here, the Celtic material culture only becomes more numerously represented from the beginning of the 2nd century BC onwards. At the same time, a greater influx of Italic imports and Roman Republican coins east of Razdrto (Ocra) is noted towards the end of the 2nd century BC. This shows that we cannot completely exclude the possibility that the vessels bearing the northern Adriatic technological signature are actually those of Italic/Roman owners, i.e. merchants and other civilians with business interests in the area of one of the Celtic kingdoms beyond the Alps, or of consular armies on expedition along the Sava corridor (Šašel Kos 2005, p. 329-334).

23The local shipbuilding practice is marked by vitality and innovativeness. In its very nature, it reflects both the significance of water transport for the local community and the interest of successive political authorities in participating in the organised transport of goods, also for military purposes. It played an important role in the formation of (supra)regional trends, perhaps even in the development of (military) barges of the “Romano-Celtic” type. Its vitality and innovativeness is clearly visible in the new technology of joining planks with densely distributed iron clamps. An autochthonous origin of this technology is supported by the substantial use of iron clamps for repairing cracks and reinforcing the bow parts on the Iron Age logboats from Vrhnika, Zakotek and perhaps also Črna vas (Gaspari et al. 2017, p. 106-107, 116-117), which is unusual in other parts of the Mediterranean and continental Europe. The Sinja Gorica barge, constructed soon after AD 3 (Erič et al. 2014, p. 218-220), is for now the earliest reliably dated representative of this new tradition, which resembles and might have been inspired by the technology of sewn ships, as could be cautiously hypothesised from the rough contemporaneity of the barges from Lipe and Vrhnika with listed examples of assembly by iron clamps from the Ljubljansko barje area. Nevertheless, the available dataset is still too slight to enable firmer conclusions in this respect. Allegedly, the reason for changing the joining technique or abandoning sewing in the construction of larger vessels in the hinterland of the north Adriatic might be connected to faster construction and easier maintenance. The chronological coincidence of the Sinja Gorica barge with the period of the Roman conquest of Illyricum and Pannonia, in which Nauportus served as an important logistic post, indicates that the new shipbuilding practice may be associated with increased supply requirements, perhaps with a military shipyard. The new method of construction of barges could have been practiced by fabri navales of northeastern Italian or even local origin, who either mastered or were at least knowledgeable of both techniques, sewing and joining with iron clamps.

24The integration of the new practice into the shipbuilding traditions of the area is also confirmed by the geographical area of distribution of the vessels with such a technique of joining the hull, all originating from the Sava-Danube river basins. Two more ships of this tradition, discovered in the Kupa riverbed near Kamensko (end of the 1st-beginning of the 3nd century; Boetto 2016, p. 1413-1416; Boetto et al. in this volume) and at Sisak (3rd century AD (?); Gaspari, Erič, Šmalcelj 2006), have features similar to the sewn ships from Vrhnika and Lipe: flat bottom with the typical transition into the sides composed of carved bilge strakes with a C- or L-shaped cross section and made of a single piece of wood, the use of vegetal luting and the prevailing use of wooden treenails for joining floor timbers or ribs and the planking.

25The feature that distinguishes the Sinja Gorica ship from both the Roman-era ships with iron clamps from the Kupa River and the sewn ships of the northern Adriatic area is metal nails for joining the frame and the planking (Erič et al. 2014, p. 242-248). In Roman-era shipbuilding, these were typically used on the barges of the Romano-Celtic tradition from the Rhône basin, the lakes of western Switzerland and Rhineland. The isolated occurrence of the use of nails within the Sava-Danube tradition could therefore be interpreted (with some reservation and pending new finds) as a workshop signature and a practice that was later abandoned. This assumption is supported by an unusual and as yet unexplained design of the vessel that is, in its researched part, marked by the absence of the usual frame – part of its functions were obviously carried out by the slim floor timbers placed into the transverse recesses in the floor planking. Judging by the similarity with the dovetail joint, the use of which reaches back to the pre-metal periods of European prehistory, and the absence of such transverse and longitudinal reinforcement of floor planking in Mediterranean shipbuilding, it is possible that this feature also relates to pre-Roman carpentry practice. All this clearly testifies to the importance of the discussed area for the innovation and transfer of naval technology at the end of prehistory and in the Early Roman period.