The Recent Find of a Cog of ca. 1240 in the Fjällbacka Archipelago, Western Sweden: A Preliminary Report

Introduction

In recent years, several well-preserved wrecks of medieval vessels have been discovered in Swedish waters. Quite surprisingly, the main bulk of these wrecks has been found along the west coast of Sweden (e.g. Gutehall et al., 2023; von Arbin, 2021a; 2022a; 2022b), and not in the Baltic Sea where preservation conditions are generally considered to be better due to the absence of the various wood boring organisms that thrive in saltier seas. As so often is the case, most discoveries were purely accidental. The recent find of the 13th-century Dyngö cog, presented in this note, makes no exception (for a discussion regarding the archaeological definition of the cog, see e.g. Crumlin-Pedersen, 2000; or Hocker, 2004, pp. 72–79). Its discovery stands out, however, in that it was found by a professional team of maritime archaeologists during the search for a completely different – and significantly younger – shipwreck.

The story goes back to 2005, when an individual contacted the regional Bohusläns Museum regarding a piece of ship timber – a large floor timber from a clinker-built vessel – which he for many years had kept under a tarpaulin in his garden. The timber had been recovered by his late father who literally had stumbled upon it whilst fly-fishing off Dyngö, an island situated in the outer Fjällbacka archipelago, approximately 100 km north of Gothenburg. The exact circumstances of its discovery and salvage are unfortunately not known, but according to the informer it was said to have been found in association with other ship elements, and supposedly in shallow water (i.e. within waders’ depth) somewhere within Dyngö’s natural harbour basin (von Arbin, 2022b, pp. 15–18).

A dendrochronological analysis of the timber provided a date of 1488 for the outermost tree-ring, using chronologies from southern Sweden. Based on this result, the likely felling date of the tree could be estimated to between ca. AD 1498 and ca. AD 1518 (H. Linderson, pers. comm., 3 March 2006). Unfortunately, at that point, Bohusläns Museum’s maritime archaeologists did not have the time or means to actively search for the shipwreck, and it was not until 2012 that a partial dive survey of the harbour was realised thanks to a collaboration between the museum and the Swedish Coast Guard. During this survey, Coast Guard divers succeeded to locate the largely buried hull of a clinker-built vessel that seemed to fit perfectly with the 2005 description. Even though the shipwreck was assumed to be identical with the previously reported wreck find, the site was never inspected by archaeologists (von Arbin, 2022b, pp. 15–18).

In September 2021, maritime archaeologists from the Department of Historical Studies at the University of Gothenburg dived the site as part of the present author’s PhD project. This ongoing project looks into different aspects of the medieval maritime transport geography of the former Norwegian province of Bohuslän, and in this context the wreck was considered to be of great potential interest. Disappointment was however big as it was soon realised that the wreck was not the one hoped for. Observations made during inspection instead suggest that this shipwreck is of a much later date – potentially even as late as from the 19th century (von Arbin, 2022c).

However, while on Dyngö, a drone aerial survey of the shallow parts of the harbour basin was conducted. In the northwestern part of the harbour, only about 120 m to the north of the just inspected wreck and close to some modern-day jetties and boathouses, an interesting anomaly could be observed in the drone footage. A brief diving inspection showed that it was indeed yet another shipwreck – but again not the one originally sought (Figures 1 and 2). The age of the new find was not immediately recognised, but certain constructional elements, such as the supporting buttresses for the keelson, indicated a possible medieval date.

Figure 1. A: Map showing the location of the Dyngö cog. B: Detailed map showing the wreck location in relation to the surrounding islands (Anders Gutehall, Visuell Arkeologi Norden).

Figure 2. Drone photo of the northern part of Dyngöhamnen. The Dyngö cog is situated in front of the two white boats in the centre of the photo (Jens Lindström, Nordic Maritime Group).

Dating and Provenance Determination

In dialogue with the County Administrative Board of Västra Götaland, the governmental body responsible for ancient sites and monuments in the region, it was decided to sample the shipwreck for radiocarbon analysis while still on the site. Naturally, radiocarbon analysis is far from optimal when it comes to determining the age of historical objects, although it has the advantage of being generally less intrusive than, for instance, dendrochronology, which requires much larger sample material. The main reason, however, for choosing this method instead of tree-ring dating was in this case the limited amount of time at disposal, which simply did not allow for more comprehensive sampling efforts at the time.

In all, two samples were collected and submitted to the Ångström Laboratory at Uppsala University for analysis. Results were available by the beginning of November 2021 (Håkansson, 2021). Sample Ua-71826, which consisted of moss, presumably used as caulking material between planks, gave a result of 926 ± 29 BP, which after calibration in the software IOSACal v0.4.1, using atmospheric data from Reimer et al. (2020), provides a date of between AD 1034 and AD 1202 (95.4% probability). Within this span, the most likely dating interval is AD 1034–1178 (92.4% probability). Similarly, sample Ua-71827, which was extracted from a small rope fragment, provided a result of 818 ± 29 BP. This gives a calibrated date of between AD 1177 and AD 1273 (95.4% probability), with a slightly higher probability (86.9%) for the interval AD 1201–1273.

As radiocarbon results suggested a shipwreck of considerable age it was considered necessary with a more precise dating of the find. Therefore, in late November 2021 the wreck site was revisited with the objective to identify and retrieve suitable wood samples for a dendrochronological analysis. In addition to sampling, supplementary measurement data were collected. Moreover, the shipwreck was photographed and filmed, the latter in order to produce a first, preliminary digital photogrammetric 3D model of the site. The site plan shown in Figure 3 is based on this model.

Figure 3. Site plan of the exposed part of the Dyngö cog, showing the presumed bow area of the vessel. The letters mark the various constructional features that have been recorded on the wreck, whereby k = keel; p = planking; f = floor timber; c = ceiling plank; b = maststep buttress, and x = crossbeam (?). Based on a preliminary photogrammetric model produced in 2021 by Jens Lindström, Nordic Maritime Group (Anders Gutehall, Visuell Arkeologi Norden).

Of the five sampled planks, all of oak, dendrochronologist Aoife Daly of the Danish company Dendro.dk was able to date four, using tree-ring chronologies from Lower Saxony (Daly, 2021). Sample Z312001_2 which contained 128 tree-rings, of which nine constituted sapwood rings, could be dated to AD 1226–1241. Sample Z3120039 only consisted of heartwood and contained 162 tree-rings. The analysis suggests that it comes from a tree felled after AD 1232. Sample Z3120059, which contained 147 tree rings, of which 22 constituted sapwood rings, could be dated to AD 1233–1240. Sample Z312007a, finally, came from a scarf or repair piece and contained 67 tree-rings, all heartwood. It is likely to derive from a tree felled after AD 1215.

If all four planks originate from trees felled at the same occasion, a common felling date of between AD 1233 and AD 1240 may be suggested (Daly, 2021). The dendrochronological dating of the vessel is thus somewhat later than indicated by the radiocarbon dated moss sample but fits well with the rope fragment. The reason for this discrepancy is not clear. That moss that had been stored for 30 years or more would have been used as caulking material seems highly unlikely. Rather, it could be expected that caulking of the vessel was renewed at some interval. Similarly, it is highly improbable that all four wood samples would come from later repairs. The most likely explanation would thus be that sample Ua-71826 for some reason was contaminated, possibly by tar used together with the caulking material (K. Håkansson, pers. comm., 16 November 2022).

During sampling, a sherd from a wheel-thrown hardfired greyware pot (Ger. harte Grauware), which typologically belongs to the late 12th until the mid-14th centuries, was found and recovered (Figure 4). In order to determine the chemical composition of the clay used for the manufacturing of the pot, a small sample was sent to OMAC Laboratories, Ireland, for ICP-MA/ES analysis. Torbjörn Brorsson of the Swedish company Ceramic Studies has compared the analytical results with reference data from analysed sites in Germany and Denmark and has been able to show that the chemical composition of the sherd shows great similarity to ceramic material from Lower Saxony and the medieval town of Bremen. Brorsson therefore concludes that the pot may have been manufactured in this area (Brorsson, 2022). Together with the timber provenance, this could suggest that the vessel was not only built, but also operated from the German North Sea coast.

Figure 4. Sherd from a wheel-thrown hardfired greyware pot found on the wreck during dendrochronological sampling (Staffan von Arbin, University of Gothenburg).

Description of the Wreck

The shipwreck is situated in approximately 2 m of water on an even silt bottom and is partially imbedded in sediment. The visible part of the coherent hull structure, which is heavily eroded and affected by wood boring organisms, measures approximately 4 × 4 m and reaches up to ca. 40 cm above seabed level (Figures 3, 5 and 6). However, by probing of the sediment, the approximate extent of the hull could be established to 10 × 5 m. Preserved elements make up the bottom section of the vessel, including keel, planking, floor timbers, ceiling planks, and buttresses. Remains are oriented in a north northwest–south southeast direction, with the presumed bow of the vessel pointing to the north-northwest, that is towards land, and the supposed fore end of the keel protruding from the sediment.

Figure 5. Part of the exposed hull section with frames, planking, and ceiling planks protruding from the sediment (Staffan von Arbin, University of Gothenburg).

Figure 6. When the water is clear, the wreck of the Dyngö cog is clearly visible from the surface. It is therefore somewhat surprising that the shipwreck was not previously known among locals (Staffan von Arbin, University of Gothenburg).

The hull appears to be heavily built and is presumably entirely of oak. The planking that is still attached to the frames is all flush-laid. Considering the dating of the vessel, this would almost certainly imply hull sides built with lapstrake planking in the typical cog manner (see e.g. Crumlin-Pedersen, 2000, p. 233). Comparisons of timber dimensions to, for instance, the Kollerup cog of around AD 1150 (Andersen, 1983, pp. 23–36), suggest an original overall length in the range of 18–20 m or more. The dimensions of the bottom planking were revealed in conjunction with dendrochronological sampling which exposed planks up to 48.5 cm wide and 4–5.5 cm thick. Tree-ring analysis shows that sampled planks have been tangentially extracted from the log, but since they are all quite heavily eroded it is difficult to tell whether they were sawn or hewn. Plank strakes appear to be joined longitudinally by means of oblique scarfs. Judging from remaining holes and impressions in the wood, scarfs were secured with iron nails with large round heads, 2.5–3 cm in diameter. Whether nails were bent or riveted is currently not known. Floor timbers are fastened to the planking with treenails that are ca. 3.5 cm in diameter.

The dendrochronological sampling revealed that the waterproofing of the hull was made in a manner typical for cogs and cog like vessels: in the plank seams, caulking in the form of moss had been inserted and was held in place by wooden laths and iron staples, commonly referred to as sintels (see e.g. Reinders, 1979, pp. 42–43). Laths, which also appear to be made of oak, have triangular cross-sections and generally measure around 3 cm wide and 1 cm thick. In order to accommodate the caulking material, the underside edges of the planks are bevelled. Retrieved plank samples show that sintel imprints are typically situated at 8–9.5 cm intervals along these edges. Unfortunately, sintels have long corroded away and their original shapes are thus not known.

An analysis by Sweden’s leading expert on mosses, Lars Hedenäs of the Swedish Museum of Natural History, of moss collected for radiocarbon analysis, identified the species as belonging to the Sphagnum genus. The material was unfortunately severely degraded, and therefore only a few of the specimens could be identified by species. These constitute Sphagnum cuspidatum, also known under its English name Feathery bogmoss. Analysis of two further samples (one from a plank seam and one from a scarf joint) which were later retrieved together with plank samples collected for tree-ring dating, identified all sampled specimens as S. cuspidatum (Hedenäs, pers. comm., 14 October 2021, and 21 February 2023). S. cuspidatum, which can be found in wetlands in large parts of Europe, has the ability to hold very large amounts of water in relation to its dry weight and was commonly used as waterproofing material in medieval shipbuilding – not least in ships of the cog type (see e.g. Deforce et al., 2014, pp. 305–306; Westerdahl, 1997).

The keelson is missing in the exposed part of the wreck but may be preserved in part underneath the sediment. Still in situ are three of its supporting buttresses – two on the presumed port side and one only partially exposed on the starboard side. To the north-northwest of these timbers is a massive beam. As it has not yet been examined in any detail it is uncertain whether it is just another buttress, or if it constitutes a crossbeam that originally may have been situated higher up in the hull. Ceiling planks, situated with some distance from one another, are resting on top of the floors. Cobbles and boulders are unevenly distributed over the site, the majority of which are likely to constitute ballast stone.

During dendrochronological sampling it was observed that bottom planks have charred inner surfaces and that sediments trapped within the hull structure contain large amounts of charcoal and other burnt residue. It is thus evident that the vessel must have been heavily exposed to fire. This is further validated by the small rope fragment retrieved for radiocarbon analysis, as well as the recovered ceramic sherd and a barrel hope discovered during dendrochronological sampling, which all showed clear evidence of having been burnt. The extent of the fire damage makes it likely that the fire, which may have been the result of either an accident or deliberate actions, was the reason the vessel originally sank on the location.

Dyngö in Archaeological and Written Sources

The island of Dyngö, easily recognisable due to its distinctive shape, is situated along the once densely trafficked medieval sailing route along the Bohuslän coast which tied together the Baltic Sea region with the western European mainland and the British Isles (see e.g von Arbin, 2021b; 2022b). On topographical grounds it may be deduced that today’s sheltered harbour, that is the bay between Dyngö and the adjacent island Trybergsholmen, known as Dyngöhamnen (‘the Dyngö harbour’), already back then must have constituted an excellent natural harbour. Not surprisingly, Dyngö is among the locations listed by the Norwegian priest and geographer Peder Claussøn Friis in his compilation of the most distinguished harbours, isthmuses, islands, and straits along the coastline between the Göta älv river mouth and the Oslo fjord, which likely was written down shortly before the year 1600 (Claussøn Friis, 1881).

Today, there are no permanent residents on the two islands. Less than 100 years ago, however, they accommodated a small, albeit lively, society. When this settlement was first established is not known. The first secure evidence of residents on Dyngö dates to the great herring fishery (ca. 1563–1589), when in the 1580s an Oslo merchant named Mickel Funtin reportedly owned a ‘salt shed’ on the island (Nyström, 1983, pp. 41–54). It was in all probability used on a seasonal basis, and the settlement likely lasted only as long as the fishery was productive. Possibly, the still undiscovered shipwreck, discussed in the introduction of this paper, could be tied to these activities. The earliest mention of a permanent settlement dates to 1659 and can be found in the first-ever Swedish tax register for Fjällbacka. According to the register of 1665, Dyngö was inhabited by pilots and fishermen. Later, in 1703, the island was formally appointed as a pilot station – a position it maintained for well over 200 years (Hansson, 1980, pp. 96–104).

Local tradition suggests that pilots resided on Dyngö already in the 13th century (Alexandersson, 2021, p. 135). The veracity of this tradition is of course difficult to assess. Recent discoveries however show that the Dyngö cog clearly not represented an isolated visit in the harbour at the time: in 2022, two green-glazed ceramic sherds, which tentatively belong to two separate jugs, were found on the northern shore of Trybergsholmen. The ceramic ware in both cases date to the 13th or 14th century and may be attributed to Yorkshire in eastern England (T. Brorsson, pers. comm., 8 October 2022). As the findspot is located approximately 200 m north of the Dyngö cog, a connection between the two seems rather unlikely. The question is thus what these sherds represent. Do they indicate a settlement or some other shore-based activity – or do they simply constitute refuse left by halting seafarers?

It is well-known that harbours that have been in frequent use for longer periods often contains extensive underwater cultural layers. Expectations were therefore high when in the spring of 2022, an archaeological survey of Dyngöhamnen was conducted by Bohusläns Museum on behalf of the County Administrative Board in response to planned water and sewer pipes on the seabed. The entire development area was inspected by diving archaeologists who also excavated a total of eight test pits along the proposed pipeline route (Widerberg & von Arbin, 2022). Disappointingly, very few artefacts were found during the survey, and none of them could be dated to prior the 17th century with certainty. It is however possible that the meagre results, at least in part, could be explained by dredging operations carried out in the central parts of the harbour in the early 20th century (see Lundberg, 1986).

Discussion

At the time of its discovery, the Dyngö cog constituted the seventh confirmed cog find in Sweden. During the spring of 2022, two further shipwrecks, discovered in Varberg, south of Gothenburg on the Swedish west coast, were identified as cogs, which gives a total of nine wrecks (Table 1). From the first ‘diagnostic’ dendrochronological results, these two vessels however appear to be approximately 100 years later than the Dyngö cog (A. Daly, pers. comm., 30 August 2022). Both wrecks were excavated, disassembled, and lifted, and ship timbers are soon to undergo digital 3D documentation (Gutehall et al., 2023). A possible tenth wreck of a cog was discovered off Nåttarö in the Stockholm archipelago in 2017. It has however not yet been examined in detail, nor has it been properly dated (Fredholm, 2021). Furthermore, in Jorefjorden, only about 4.5 km east of Dyngö, the wreck of a vessel with several distinct cog features, dated to the early 1260s, have been partially excavated (von Arbin, 2022b; in prep.).

Table 1. Finds of confirmed cogs in Sweden until 2022 listed in chronological order.

Name	Location	Date	Wood provenance	Reference(s)
Kuggmaren I	Stockholm archipelago, E Sweden	1215	Denmark	Adams & Rönnby, 2002; Hocker & Daly, 2006
Dyngö III	Fjällbacka archipelago, NW Sweden	1233–1240	Lower Saxony, NW Germany	Daly, 2021; von Arbin, 2022b, and this publication
Kronholmen	Västergarn, Gotland	1245–1251	Kalmar region, incl. Öland/Gotland, SE Sweden	Rönnby & Björdal, 2023
Bossholmen	N Oskarshamn, SE Sweden	After ca. 1270	East Skåne, Sweden	Daly, 2009
Helgeandsholmen II	Central Stockholm, E Sweden	ca. 1330	Not known	Varenius, 1989
Varberg 1	Central Varberg, W Sweden	After 1346	Netherlands/Belgium/NE France	Daly, pers. comm. 30 August 2022; Gutehall et al., 2023
Varberg 2	Central Varberg, W Sweden	1355–1357	N Poland	Daly, pers. comm. 30 August 2022; Gutehall et al., 2023
Mollösund II	Orust, NW Sweden	ca. 1365	Funen/Zealand/W Denmark/N Germany	von Arbin & Daly, 2012
Skanör	Skanör, S Sweden	After 1382	SE Baltic	Ossowski et al., 2003

The number of recorded wrecks in Europe that meets the archaeological definition of a cog, as formulated by, among others Crumlin-Pedersen (2000) and Hocker (2004), is constantly increasing, and currently the figure probably approaches 40. With a dating to between AD 1233 and 1240, the Dyngö cog constitutes one of the earliest examples of a vessel built in this manner. Other early cog finds include the vessels from Kollerup (ca. 1150); Kolding (1188–1189); Skagen (ca. 1195); Kuggmaren (1215), and Kronholmen (1245–1251). Dendrochronological analyses have shown that the timber origin for the first four wrecks is southern Denmark (Hocker & Daly, 2006), whereas the fifth seems to originate in southeastern Sweden (Rönnby & Björdal, 2023). This is highly interesting considering that the emergence of this ship type in much of the previous research so strongly has been associated with the German Hanse. The Dyngö cog is thus the first known cog that appear to have been built within what would later become the core area of the Hanseatic League.

However, as shown by previous research, and most recently by the impressive work of Vlierman (2021), the medieval cog is a theme with many variations. Not surprisingly, early vessels seem to display a larger variety in technical solutions than later ones, which appear to be much more uniform in terms of construction. Early specimens, such as the Dyngö cog, is thus hugely important for our understanding of the origin and subsequent technical development of this particular ship type. Further investigation and recording of the Dyngö cog would thus be highly motivated from a ship archaeological point of view. Although the vessel is severely burnt, there may still be good chances for finding well-preserved arte- and ecofactual material within the hull, as suggested by ceramics and barrel parts observed on the site. An in-depth study of the wreck may thus not only provide significant data on 13th-century ship construction, but also on European maritime trade during this period.

Statement on Permissions

Sampling of the Dyngö cog for radiocarbon and dendrochronological analysis was conducted under permits Nos. 431-35395-2021 and 431-49881-2021, issued by the County Administrative Board of Västra Götaland.

Acknowledgements

I would like to thank Aoife Daly, Dendro.dk, who conducted the dendrochronological analysis of the Dyngö cog; Lars Hedenäs, the Swedish Museum of Natural History, for his analysis of caulking mosses, and Torbjörn Brorsson, Ceramic Studies, who arranged for the ICP-MA/ES analysis of the ceramic sherd and carried out the subsequent provenance determination. Many thanks are also due to Jens Lindström, Nordic Maritime Group, and Anders Gutehall, Visuell Arkeologi Norden, for their participation in the 2021 fieldwork as well as for their continuous support. I am equally indebted to Anders Säldemark, Dyk-Leif, for special arrangements in conjunction with fieldwork, and Anders and Lena Ekelöf, for their hospitality and practical assistance. Last, but not least, the two anonymous reviewers are thanked for the constructive comments on the paper.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

Funding of the fieldwork and the various analyses were granted by Adlerbertska forskningsstiftelsen and VOTO – Voice of the Ocean.