1On 3 October 2016, a local resident reported a previously unknown shipwreck at Hörnum Odde on Sylt’s southern tip to the State Archaeology Department of Schleswig-Holstein (ALSH). The substantial hull remains came to light as a result of strong tidal currents and an underlying erosion process. The site was located at the outer extent of the intertidal zone and only accessible at low tide in offshore winds. Thus, the location of the site imposed a special challenge as its location in the surf zone permitted neither an underwater survey, nor a proper terrestrial excavation. Therefore, the recording method had to be extremely time-efficient in order to gather as much data as possible within a very short time frame.
2In view of the challenges described above, the first investigation was scheduled for 6 October when low tide coincided with offshore winds. This provided optimal working conditions, as the wreck remains were clear of the water for about an hour. The wreck’s outline was measured with a DGPS device on the basis of SAPOS, a German nationwide satellite-based reference networked service, which decreases the margin of error in real time surveying, reducing deviations to a maximum of 10-20 cm for the device used by the ALSH. Over 200 photos were taken, to be used later for Structure from Motion (SfM) photogrammetry. Several test pits were dug to find the keel, but to no avail. A number of samples were taken from both the planks and the frames for later dendrochronological analysis, and a piece of a plank was salvaged for detailed ex situ documentation. The stormy season made it difficult to find an adequate date for a follow-up investigation in which all prerequisites were met, i.e. low tide around midday, offshore winds and availability of field staff. In the meantime the ALSH received reports and photos of the wreck site from its established local contacts. Based on this information it became clear that the wreck site was subject to considerable mechanical destruction whenever the site was turned into a surf zone with each incoming tide. Thus, it was decided not to wait any longer for optimum conditions and to visit the site again on 9 December, when a different section of the wreck could be observed, while the previously investigated section was no longer visible, having been washed away. Due to unfavourable onshore winds, the wreck site remained inundated at low tide and so photographic documentation suited for an SfM model was out of question. The wreck was only accessible by wading into the water, which permitted only SAPOS surveying and manual recording. Only two pieces of timber could be sampled at this time. Although the circumstances of the December investigation were very unfavourable, the effort was worthwhile, since by the New Year the entire wreck site was reported missing. It was reclaimed by the sea as a result of coastal erosion and winter storms.
3The post-excavation involved georeferencing both wreck sections with ArcGIS, drawing a wreck plan on the basis of the SfM-generated orthophoto, observations from photo details as well as the in situ manual recording of the inundated wreck section. The desk-based assessment of the geomorphological changes, the wreck’s construction details, the drawing of the retrieved plank fragment, as well as research carried out to contextualise the wreck took up the greatest portion of the time. An archival study of recorded historical strandings in search of the vessel’s identity has yielded no match so far.
4Upon georeferencing the wreck site it came as something of a surprise that the site was evidently situated in the middle of the Hörnum Odde dune (fig. 1). As it turned out, the fairly recent update of the digital topographic map from 2014 was not recent enough to keep up with the dramatic changes to Sylt’s coastline. A comparison with a more recent aerial orthophoto taken in July 2016 confirmed that the location of the wreck in relation to the shoreline was indeed accurate and not the result of a measuring error. By July the dune edge had advanced close above the wreck site and by October it had moved already some 50 m further inland beyond the wreck site.
Fig. 1: The find location of the Hörnum Odde wreck (red dot) in the regional context (top left) and georeferenced in the digital topographic map of 2014 (background)
As the historical coastline of 1793 (mean sea water level) and the dune edges of 1878, 1929, 1978, 2014 and 2016 indicate, Sylt’s Odde has been subject to erosion along its western shore and sand aggradation at its southern spit for centuries. A shift towards a more southeasterly deposition of sediments in more recent times is reflected by geomorphological data, as indicated by the different shades of blue (= erosion) and red (= sedimentation) for the period of 1982-2012 with the total values of depth differences. This underlying process explains the rapid land loss at the wreck site itself. While the dune edge has advanced closely above the wreck site by July 2016, it has moved considerably further inland at the time of the discovery of the wreck in early October of the same year, when the photo (bottom left) was taken.
(graphics D. Zwick; photograph ALSH / DTK25-map LKN-SH; geomorpological data GDI-BSH)
5The great coastal erosion process on Sylt’s western shoreline is not a new phenomenon. It is an ongoing process that has taken place for centuries if compared to historical coastlines (Ahrendt, Köster 1998, p. 39). The gradual erosion process along the western coast goes hand in hand with a sedimentation process to the south, forming a sandy spit – the Odde. When compared to a long-term 30-year geomorphological model based on depth measurements between the years 1982 and 2012, the erosion of the western coastline has remained a constant feature with a shift of the sedimentation process to the east. Between the years 2013 and 2016 some 30 to 60 m of coastline was lost annually due to this process (LKN-SH), evidently reaching full peak in 2016 as the maximum annual value of 60 m corresponds to the erosion of 10 m as evidenced by the gap between the two wreck sections visited in a two month interval (fig. 2).
Fig. 2: The wreck site as recorded on 6 October and 9 December 2016
The dotted line indicates the position of the missing keel. An SfM model (bottom right) was generated for wreck section 1, while wreck section 2 (inundated at the time of the in situ documentation) is based on a hand sketch, in which the treenail arrangement was recorded for two exemplary segments. The locating surface may indicate the position of a bulkhead.
(graphics D. Zwick; photograph ALSH; SfM-model J. Enzmann, F. Wilkes)
6The geomorphological process explains the location of the wreck in the middle of the dune. Based on this data alone, one could infer that the shipwreck could not have occurred much later than 1793, providing a rough terminus ante quem.
7The unearthed portion of the wreck covered an area of circa 16 by 5 m (fig. 2). The wreck was situated in a north-northeasterly direction, with loose hood ends observable in the southern part (section 1) indicating one of the ship’s extremities, tentatively addressed as the stern section. The northern part (section 2) corresponds to the midship section and continued beneath the sandy beach. The full length of the wreck site remains unknown.
8The floor timbers are not interconnected with futtocks and are only fastened to the planks with wedged treenails of 35 mm as the most common diameter. In the chine area, floor timbers and futtocks form a parallel, alternating pattern. Some of the observed futtocks display an articulated chamfered edge shortly above the turn of the bilge, so these could be more appropriately described as bilge futtocks, known by Dutch shipbuilders as zitters (Hoving 2012, p. 64). The scantlings of the frames are quite modest, with moulded dimensions of around 15 cm and sided dimensions of around 30 cm in the amidships section (section 2), gradually decreasing to 20 cm +/- 5 cm for the V-frames in the stern section (section 1). The modest scantlings are contrasted by the massive dimensions of the oak planks, with widths between 24-35 cm and thicknesses of 6-7 cm. Similarly, the ceiling planks of pine have great widths of 35-38 cm and a regular thickness of 3 cm. The planks of a strake were connected with flat scarfs with stopwaters. The retrieved plank fragment (sample 3, figs. 2-3) contained a multitude of further diagnostic features. One was the observation of rectangular wooden plugs of pine, which were referred to as spijkerpennen by the contemporary Cornelis van Yk (1697, p. 41). Spijkerpennen are diagnostic for the so-called “Dutch flush” construction method, which occurs in connection with the abovementioned irregular, modestly dimensioned, non-interconnected framing system (Maarleverld 1992, 2013). This method is a shell-first construction method with flush-laid planks: a “carvel” in appearance only. The flush-laid planking is temporarily held together by cleats nailed onto the planks from both sides of the hull (fig. 4), only to be removed later once the framing has been inserted, with the nail holes eventually plugged with spijkerpennen. Similar to clinker constructions, other devices were used to control the run of the strakes, such as large planking tongs, referred to as boeitang by Nicolaes Witsen (1671, p. 169). The fact that there is no discrete term for this tong in a seafaring nation like England, as reflected by the circumscription “a sort of pincers to join the planks” in a Dutch-English dictionary (Sewel, Buys 1766, p. 127), highlights the uniqueness of this device in Dutch shipbuilding. Further characteristics also point to a Dutch construction, like what appears to be the charred inner surface of a plank (fig. 2, S3) beneath the frames. Despite the massive thickness of the oak planks, they had to be bent, and the process of making planks more malleable by charring the surface was also described by Nicolaes Witsen (Witsen 1671, p. 207; Hoving 2012, p. 23). Charring can also be seen as a typical feature of Dutch 17th-century shipbuilding: it was later replaced by steaming.
Fig. 3: Drawing of a plank fragment (sample 3) retrieved from the garboard strake, with the tapering edge of the scarf visible
This piece features a number of diagnostic characteristics like the spijkerpennen (blue), treenail holes, rectangular iron nail shafts (red) and nail head impressions (dotted line around the nail shafts), and dowels (brown) in the patch, which was fastened with four iron nails. The plank sample was sawn through at the patch: its cross section revealed that the inner recess had a conic plane, probably due to the difficulty in applying the tool horizontally in this restricted space. As indicated by the knot on the interior, the insertion of the patch over the same on the exterior can be ascribed to an effort to make the vessel watertight. Moreover, residues of caulking material were present on the exterior surface of the plank, fused in by an iron concretion. All measurements and diameters are provided in mm.
(graphic D. Zwick; photograph ALSH)
Fig. 4: Demonstration of the Dutch flush method, where the flush-laid shell was assembled prior to the erection of the skeleton
Temporary cleats held the shell together, which were removed once the frames were inserted and their nail holes plugged with spijkerpennen. As Nicolaes Witsen demonstrated (top left), not only cleats (l) were used but also boeitangs (e) and other aids.
(drawing after Hoving 2012, fig. 2.55 courtesy of G.A. de Weerdt; top left, after Witsen 1691, p. 168, modified D. Zwick)
9The choice of wood species is of particular interest. While the framing timber and planks of the stern section are almost entirely of oak, with the exception of the pine ceiling, the frames in the amidships section are all of pine. This is a clear indication that this is the wreck of a merchantman, since portions of the hull which were under less stress were often completed with a lighter coniferous species to reduce the deadweight of the hull, thereby increasing its cargo carrying capacity (Hoving 2012, p. 22).
10The stern section features four strakes of planking that appear to be doubled. Although this could be interpreted as a repair or later addition, the impression of a vessel built with two layers of planking was further corroborated by truncated treenails, which were particularly apparent in the midship section (fig. 2, wreck section 2). The reason for the truncation was described by Thijs Maarleveld (1994, p. 153) in the context of the building sequence of what he dubbed the “Double Dutch” method: a first row of treenail holes was bored to connect the planks with the inserted frames, then a second layer of planks added and fastened once again with an additional set of treenails, now protruding through all layers of the hull; the first and the second layer of planks, the frame and the ceiling planks. In several cases the new treenail holes truncated the earlier treenail insertions. Another feature of the “Double Dutch” method is the ample use of waterproofing material applied broadly between the first and the second layer of planking (Maarleveld 1994, p. 153), which, according to a contemporary Dutch source, contained wax and resinous matter onto which a matting of animal hair was applied as filling material before the second layer of planking was added (van Yk 1697, p. 91). On the outer side of the retrieved plank (fig. 3), probably the garboard of the first layer of planking, were indeed traces of waterproofing material identified as animal hair, which survived thanks to an iron nail concretion. An effort to make the inner planking watertight is also reflected by a patch inserted over a knot.
11In early December a local resident sent photos of the keelson to the ALSH. A few days later, on 9 December, the keelson had already gone missing, so all observations could only be inferred from the photos. The keelson measured approximately 50 cm in width and had an integrated mast step, with a carved diagonal cross at its base with a protruding hole in the centre, probably a drain. The position of the keelson and mast step could be reconstructed precisely by a protruding treenail observed on 9 December and thus confidently integrated into the general plan (fig. 2). The mast step featured chop marks and the wreck had evidently been scrapped. The scrapping of wrecks was a welcome source of income for the local population and ducal edicts specified shares and penalties to deter beachcombers and illegal wreckers (Kühn 1999, p. 10–11, 29). According to the Gottorfische Strandordnung, a ducal edict of 1712, stranded crews had only three tides to get their vessel either afloat again or to save their belongings before the wreck site was cleared for scrapping and salvage (Englert 1997, p. 47–48).
12The following results are based on the analyses carried out by Dr Karl-Uwe Heußner of the German Archaeological Institute in Berlin. Two samples taken from oak floor timbers (fig. 2, S1, S2) contained sapwood and date to 1684 and 1690. Another two samples taken from oak planks (fig. 2, S3, S5) without sapwood have the years 1647 and 1649 as last tree rings, respectively. The chronological discrepancy can be explained by the absence of sapwood rings and waste of material, which is common for planks due to their pre-specified widths. Thus, 1690 is the terminus post quem, indicating a ship construction date in the same year or shortly thereafter.
13The framing timber originated from the central Havel or Prignitz region in Brandenburg. Given that there are separate reference curves for both regions, the modest t-value of 5.7 should not be underrated (Karl-Uwe Heußner, personal communication). The planks correspond to timber chronologies in northern Poland. The pine timber of the amidships section was unfortunately not accessible for sampling, but it could be surmised that it would have added a third discrete provenance.
14The regional discrepancy of provenances is not surprising, because different construction elements necessitated timber products of various species, qualities and properties. These were commonly imported by citizens of the Republic of the Seven United Netherlands, which was almost entirely deforested and yet operated the largest merchant fleet in the world at that time. Framing timber from Brandenburg would have been imported as “Hamburg timber”, since Hamburg was the point of transshipment for oak timber cut in Brandenburg and floated down the Havel and Elbe rivers. At that time, Hamburg was known as one of the most important ports for acquiring compass timber (Hoving 2012, p. 22), probably because “Hamburg timber” was appreciated as particularly resistant and durable due to its high content of heartwood (Karl-Uwe Heußner, personal communication). The Dutch were amongst the main importers of “Hamburg timber” and became very engaged in the oak timber trade in Brandenburg from 1650 (van Tussenbroek 2006). The planks were most likely transshipped via Danzig (Gdańsk), which remained one of the most important Baltic Sea ports for timber exports next to Königsberg (Kaliningrad) and Courland in the 17th century (Bonde et al. 1997, p. 203).
15Although the Hörnum Odde wreck has several constructional features in common with other 17th-century wrecks from the North Frisian Wadden Sea (Zwick, Klooß 2018, p. 212–213), which was an area under heavy Dutch influence, its most striking feature is the “Double Dutch” construction. The earliest known wreck featuring such construction is the Scheurrak SO1 of circa 1590 from the Netherlands, thus, with the Hörnum Odde wreck of 1690, the longevity of the “Double Dutch” phenomenon can be attested for a century. Aside from the Scheurrak SO1, most “Double Dutch” wrecks hitherto identified were owned by the Vereenigde Oost-Indische Compagnie (VOC), or Dutch East India Company, which usually had an additional third layer of pine as sacrificial planking. These wrecks are the Mauritius (1594/1602), Batavia (1628), Avondster (circa 1641), the Angra C wreck (17th century) and the B&W 2 wreck (1606); the latter being a Dutch-built Danish East Indiaman (van Duivenvoorde 2015a, p. 147). There can be no doubt that the “Double Dutch” technique was “phenotypical” for the VOC from 1602. This is reflected in a 1643 letter addressed to the VOC Chamber of Enkhuizen, which sought to reduce shipyard maintenance time in the colonies and therefore recommended an addition of pine planks to protect oak planks from shipworm (van Duivenvoorde 2012, p. 247).
16Based on the comparative evidence, an interpretation of the Hörnum Odde wreck as a VOC ship seems possible. Its find location is not problematic, as VOC ships often chose a detour through the North Sea circumnavigating the British Isles as evidenced by VOC wrecks like the Kennemerland, sunk in 1664 off the Shetland Islands (Muckelroy 1976), or the Adelaar, wrecked in 1728 off the Outer Hebrides (Martin 2005). Particularly during the Nine Years’ War (1688–1697), in which the Dutch Republic was part of an alliance against France, VOC ships avoided the English Channel in fear of French privateers. In 1690 alone, two returning VOC ships wrecked off the Shetland Islands (Bruijn et al. 1979b, p. 133). There is also a historical record of a VOC fluyt, which foundered off the coast of Sylt, but its date does not match with that of the Hörnum wreck. It was the VOC fluyt Amstelland, built in 1750, set sail from Texel on 7 September 1751, but ran ashore on Sylt 11 days later (Bruijn et al. 1979a, p. 534).
17However, a VOC association is not the only possible interpretation, as a historical document of 1660 mentions that Dutch whalers sailing to Spitsbergen also fitted out their ships with two layers of oak planking (van Duivenvoorde 2015b, p. 359). An additional sheathing would have made sense not only against marine borers, but also as structural reinforcement against ice floes.
18Last but not least, such reinforcement would have also been beneficial when beaching ships on a tidal shore to stevedore at low tide, which was a common practice in the Wadden Sea area. On the Danish west coast such vessels were called sandskuder and were also often double-planked, but in contrast to “Double Dutch” constructions, sandskuders received the second layer of planking not from the beginning, but during a later rebuild after the hull had worn out and become leaky (Gøthche 1985, p. 304). Dutch fishermen kept the practice of beaching ships alive well into the 19th century, and the process inspired the British painter Edward William Cooke to numerous paintings in which the pincks of Scheveningen served as a common motif. Thus, the Hörnum wreck could also be interpreted in a very regional context and may not even necessarily reflect an accidentally stranded wreck, but possibly a vessel that fell dry at low tide but could not get afloat again due to a strong onshore gale, or the influence of a neap tide, while the hull was settling deeper into the sediments with each missed tide, until it was cleared for salvage.
19Despite the adverse site conditions, a number of significant details could be observed and documented, which allowed an unequivocal identification of the shipbuilding tradition fulfilling all the criteria of “Dutch flush” and “Double Dutch”. The question of the ship type or even the ship’s actual identity can not yet be clarified, but the dimensions of the vessel certainly suggest that this must have been a multi-masted ship, perhaps a fluyt or pinnace. At the conference, one participant advocated to drop the term “Double Dutch” in view of doubleplanked vessels of other shipbuilding traditions, but this author believes that “Double Dutch” remains a useful term to refer specifically to doubleplanked vessels built in the idiosyncratic Dutch shell-first method.
20
I wish to thank my colleague Dr Stefanie Klooß for the organisation of the fieldwork and for tasking me with the documentation, research and publication of this fascinating wreck. I also wish to thank my other ALSH colleagues, Jan Fischer, Linda Hermannsen and Heiner Menzel for the pleasant cooperation. I am grateful to Dr Karl-Uwe Heußner for the swift dendrochronological analysis and for providing further comments, and to Jonas Enzmann and Feiko Wilkes for generating an SfM model with Agisoft Photoscan. Karsten Bracker, Dr Anton Englert and particularly Dr Wendy van Duivenvoorde provided interesting details and literature, which helped to contextualise this wreck.