1The Oseberg ship was built in western Norway in AD 820 and was placed in a burial mound near Tønsberg, Norway, in AD 834 (Bonde 1994, p. 138-141; Bonde, Stylegar 2009, p. 163-164). The 21.5 m-long and 5.1 m-wide ship is the earliest known Viking ship with traces of sail and rigging (Christensen 1997, p. 302). It is therefore an important source in terms of understanding the development and composition of rigging in Viking Age Scandinavia, with a view to exploring the ship’s sailing performance (Bischoff 2017, p. 1-5). The ship has previously been interpreted as merely a royal yacht intended for inshore sailing in fair weather (Brøgger, Shetelig 1951, p. 154-156; Christensen et al. 1993, p. 149-150; Christensen 1996, p. 83), however, given the locations of its site of construction and subsequent deposition, the ship had obviously sailed in open waters in the North Atlantic, the North Sea, and Skagerrak, and is therefore likely to have performed well under sail.

2To date, two full-scale reconstructions of the Oseberg ship have been built. The first, Dronningen, (The Queen) capsized and sank during the first sea trials in 1988. The second, Saga Oseberg, was launched in 2012 and sailed badly until a change of methodology gave very different results. This gave rise to an awareness of what questions to ask – and how to ask them – when test sailing a full-scale reconstruction.

3Dronningen left a long list of questions in its wake when it capsized (Andersen, Andersen 1989, p. 246-253; Carver 1995, p. 289-304). The full-scale reconstruction was based on drawings of the exhibited ship, so new studies of the archaeological ship were therefore carried out in 2007. The new reconstruction led to a hull form with more buoyancy, broader bow and greater keel rocker (Bischoff 2012, p. 337-342). A hydrodynamic test carried out in 2008 showed that the new hull form had more stability and improved performance (Prince et al. 2008; Bischoff 2012, p. 341-342).

4The new hull form resulted in the second full-scale reconstruction of the Oseberg ship, Saga Oseberg, which was launched in 2012.

5Saga Oseberg was built as close to the original ship as was possible, using the same methods, materials and tools as on the original ship (Finderup 2018). It was therefore expected to perform close to what the original ship did. The sail and rigging was reconstructed based on traces of the rigging in the archaeological ship. In the Oseberg ship the mast position is preserved in the keelson and the mast partner, and the tack and sheet positions can be deduced from holes and clamps. These traces of the rigging give a distinct frame for the approximate width of the square sail that the ship originally carried (Bischoff 2017, p. 3-12). Based on iconographic and written sources from AD 700 to AD 1300, Saga Oseberg was reconstructed with a right-angled square sail of handwoven wool of the same quality as a fragment found in the original ship (Andersen 1995, p. 249-270; Andersen, Bischoff 2016, p. 137-156; Bischoff 2017, p. 5) (fig. 1)

6Before it was launched, the new reconstruction had already been subject to criticism from traditional Norwegian boatbuilders, with doubts being raised as to the validity of the new hull form, especially its considerably rockered keel. Test sailing the full-scale reconstruction would now show if the changes in the hull form had led to an improvement in sailing performance.

1. Methodology of test sailing No. 1

7The first test sailing of Saga Oseberg was undertaken in Norway in 2013 using the principles employed for handling traditional square-rigged boats from western Norway, more specifically the Nordland boat and the Åfjord boat (Eldjarn, Godal 1988, p. 47-67, 1990, p. 256-299) (fig. 2).

8It seemed reasonable to do so because there are striking parallels between these ethnographic boats from the 19^th century and ship finds from the Viking Age in terms of the hull ­construction and composition of the rig (Andersen et al. 1997, p. 145-151; Andersen, Andersen 1989, p. 53-136; Crumlin-Pedersen 1997, p. 130-140). This has led to the accepted understanding that these boats had developed in a direct line from the Viking Age.

9Even though the Nordland boat and Oseberg ship have similarities, the ships also have differences, and these differences might be of greater importance than expected. The Nordland boat has a trapezoidal sail, straight keel, stretched bottom and sides, and a stern rudder, while the Oseberg ship has a rectangular sail, curved keel, curved side and bottom, and a side rudder. The mast’s angle and position in the Oseberg ship was fixed in a keelson and mast partner, while the Nordland boat has an adjustable mast, in the sense that the mast step can be moved, and the heel angle of the mast can be changed: all significant details when it comes to the handling of the boats.

10The principles of the test sailing procedure of the Nordland boat were written down in the 1980s, based on interviews with old fishermen, who had sailed these boats for a lifetime (Eldjarn, Godal 1990, p. 68-69; Tordsson 1999, p. 66-72). According to the procedure, the ship should be trimmed 1-1.5 degrees aft, which is considered as being the normal trim condition. After this, the ship is tested under oars in side wind to check if the hull, with the sail lowered, has a tendency to either a lee helm or weather helm. If needed, the ballast is moved longitudinally in the hull, so that the hull moves straight through the water. After this, the ballast position cannot be changed (fig. 3).

2. Trim and balance

11A test sailing inevitably includes theorising about hydro- and aerodynamic forces, and how these affect the ship’s trim. This can be briefly explained as follows.

12The ship is in trim when it manoeuvres in a balanced manner through the water without lee helm or weather helm. This requires balance between the sail and the hull. When the aerodynamic centre in the sail and the hydrodynamic centre on the hull are vertically aligned, the hull and sail are in balance and the ship in trim. If not, the ship will either have lee helm or weather helm. The balance can be influenced by positioning the sail differently, or trimming the hull with ballast. If the ship is trimmed down in the aft or down in the bow, the ship will turn either away from the wind or towards the wind respectively. The same applies to the sail. If the sail is moved aft, by changing the tack position aft, the ship will have weather helm, or lee helm if done the opposite.

3. Test sailing No. 1 in action

13The first test sailing was undertaken by a full crew of experienced sailors – including the author – used to sailing traditional Norwegian square-rigged boats, under the direction of a Norwegian traditional boat researcher, Jon Godal.

14The preconception of Godal prior to the test sail was that the Viking ship was expected to behave like the traditional Norwegian boat (Godal 2013, p. 1). Therefore, the stone ballast was placed in Saga Oseberg giving a trim at one degree aft, in accordance with traditional boat handling. The following rowing test in side wind showed neither lee nor weather helm on the hull, and the ballast was therefore kept in the tested position, with 2,500 kg in the central area by the mast, and 500 kg in the very aft of the hull.

15When the sail was set at the tack position for beating, it quickly became apparent that the ship manoeuvred badly. The ship was unable to tack, and drifted sideways on the course. Since the hull form of Saga Oseberg was more rockered than the exhibited Oseberg ship, Godal considered the rudder to be too short and thus it did not go deep enough into the water under the keel. In an attempt to improve the ship’s steering, the rudder was replaced with a larger and longer rudder, disregarding the fact that the Oseberg ship was found with the original rudder in place. There was therefore much discussion as to whether or not such a deviation from the archaeological material should be allowed when conducting experimental archaeological research. According to Godal, the test sailor cannot be expected to relate to the archaeological material, but only to the full scale reconstruction as it is built: and the test sailor’s job is to make the ship function. If the ship needs changes in order to function, this is a direct message to the ‘reconstructor’ that the reconstruction is wrong (Godal 2013, p. 1, 2014, p. 1-2).

16In any case, the increased rudder size was not enough to steer the ship. Saga Oseberg was still not able to beat to windward, due to the substantial lee helm. It was not possible to obtain balance between the sail and the hull.

17Saga Oseberg’s poor sailing performance was taken as proof that the reconstruction of the hull form was incorrect. Godal presented a series of conclusions, which he saw as explaining the poor performance (Godal 2013, p. 7-8):

• As expected, the rockered keel gave the ship an unstable trim;
• It was necessary to lengthen the rudder because of the keel rocker. This indicates that the keel was not rockered originally;
• The sail was too broad or not trapezoidal enough to balance with the hull, because the keel’s operational area is too short due to the rocker;
• The ship had lee helm and was unable to beat to windward. Again, this was interpreted as being a result of the curved keel’s short operational area;
• The ship is not seaworthy when beating and rowing is not possible.
• Godal made some suggestions as to how to experiment with improving Saga Oseberg’s performance (Godal 2013, p. 8-9):
• Lengthen the rudder by up to 50 cm more than the original rudder;
• Make the keel straight on the mid-part by adding a false keel;
• Let the bow and stern drop down, to make the keel curve straighter;
• Heel the mast aft in order to move the aerodynamic centre aft to balance with the hull;
• Redesign the sail to be narrower and trapezoidal.

4. Discussion of test sail No. 1

18Roughly speaking, these suggestions closely resemble a Nordland boat. Godal did not put aside his own preconceptions, and instead based his conclusions on his knowledge of ­construction and experience of sailing the Nordland boat. This made him question the rockered keel throughout the whole test sailing process.

19The testing of the full-scale reconstruction is not only a direct message to the constructor but is also a direct message to the test sailor. However, Godal’s suggestions listed above all focus on changing the reconstruction, instead of the boat handling.

20The positioning of the ballast, which trimmed the ship one degree aft, had influenced the trim by moving the hydro­dynamic centre of the hull aft. This is the reason why Saga Oseberg would manœuver better with the mast heeled aft, because by doing so, the aerodynamic centre in the sail also moved further aft and corresponded with the hydrodynamic centre. Making the sail narrower and trapezoidal would have had the same effect, since the aerodynamic centre lies in the front 1/3 of the sail (Andersen, Andersen 1989, p. 127).

21From an experimental archaeological point of view, the mast position as well as sail breadth and shape cannot be changed since they are preserved in the original ship. The mast position and angle in the Oseberg ship is fixed between the keelson and the mast partner supporting the mast. It was not meant to be – and cannot be – in any other position or angle. The sail breadth is the ship’s most probable original sail breadth. The iconographic material shows that Viking Age sails were right-angled and not trapezoidal, as those used on boats from the North Atlantic area in more recent times (Bohlmann 2014, p. 113-119; Bischoff 2017, p. 12-22).

5. Methodology and action of test sailing No. 2.

22A further series of test sails was undertaken on Roskilde Fjord in 2015 under the direction of the author and using a different methodological approach. The hull, sail and rudder functioned in unison on the original ship and therefore should be able to again without changing their form. In order to work backwards from the archaeological evidence to the possible boat handling of the ship in its own time and context, the ori­ginal mast, sheet and tack position must be kept when test sailing. Instead of the suggested changes to the archaeological evidence (hull, rudder and sail), boat handling was the primary focus of the second test sailing program.

23Based on the firm belief that the archaeological ship embodied experience and knowledge from the 9th century, the rudder was changed back to a reconstruction of the Oseberg ship’s original rudder. The reconstructed rudder functioned better with the ship but did not solve the overall problems of the ship’s manoeuvrability.

24The investigation of Saga Oseberg’s ability to beat to windward continued. It was clear that the ship had a considerable lee helm with close-hauled sheets, and beating and tacking was almost impossible.

25In order to find the right trim of the ship, “off-duty” crewmembers formed a group and moved metre by metre longitudinally in the ship, until the ship had a stable course with the rudder in neutral position. The group of crewmembers (700-800 kg) was standing two metres in front of the mast when the hull and sail finally balanced. The ship’s trim was then mea­sured as 0 degree.

26All crewmembers moved to where they would normally be positioned during sailing, and the ballast was adjusted. The 500 kg that was placed far aft in the ship by the rudder during the first test sail – in order to trim the aft down 1 degree – is a tough way of trimming a ship, because the weight arm from the ship’s centre of buoyancy is considerable, and diminishes the crew’s ability to trim the ship longitudinally with their body weight.

27In order to place the ballast so that the hull trimmed at 0 degrees, the 500 kg was moved forward to the centre of the ship. In the Oseberg ship, the deck planks were fastened with treenails, except in the centre area around the mast, indicating that the ballast was originally positioned here. The new balance between hull and sail resulted in a very different sailing performance. Suddenly Saga Oseberg was able to beat. Tacking and manoeuvring went easily and the boat handling was less heavy for the crew (fig. 3).

28All the oars from the Oseberg ship were found complete, but ever since the excavation they have been interpreted as being too short for the ship, compared to more recent oars for open boats. The length of the Oseberg oars varies up to half a metre, so the shorter oars could have been used amidships and the longer oars in the bow and stern, which rise higher above the water. The oars for Saga Oseberg were made equivalent to the originals and proved to match the longitudinal curved shape of Saga Oseberg due to their individual lengths, and rowing showed that it was possible to row the ship at a speed of 5.0 knots.

6. Conclusion

29The two test sailing methods gave different results because they were based on different approaches. By choosing a deli­berate approach in test sail no. 2 and accepting the archaeo­logical ship, its construction and design as the base for the investigation of the full-scale reconstruction, the second test sailing led to a marked improvement in the ship’s sailing capabilities, and provided information about the capabilities of the Oseberg ship from 820 AD, which was the purpose of the experimental archaeology project.

30This shows that the methodology chosen to test a full-scale reconstruction is vital and underlines the need for developing a set of principles for the testing of full-scale reconstructions of archaeological ships, in order to avoid testing from a contemporary, and potentially subjective, perspective.

31Prerequisites for the test sailor:

• Experience with trim and balance between the hull and the sail;
• An open and reflective approach towards hull, sail and trim;
• Part of an interdisciplinary team;
• Knowledge of the archaeological foundation;
• Experience with an equivalent type of ship;
• Experience with equivalent rigging;
• The key tenets of the methodology:
• The testing follows the experimental archaeological approach defined for the project;
• Focus is kept on the purpose/aim of the project;
• The ship’s details are used analytically to deduce how the ship was originally sailed;
• The original traces of the rigging set the frame for testing;
• Experience is used for reflection and for asking relevant questions about ship and boat handling;
• Presumption is always there: its influence on the given problem must be recognised;
• With the benefit of hindsight, several indications of the Oseberg ship’s original trim can be observed on the ship itself. When obtaining balance between hull and sail at 0 degrees, the ship handed down the following characteristics from its former use – a direct greeting from the past:
• The ballast is placed in the only open area in the ship’s mid-part;
• The mast, which is fixed between keelson and mast partner, stands vertically;
• The ends of the stem ornamentations are right above the waterline both fore and aft (fig. 4);
• The individual length of the original oars fits the distance from oar holes to the surface;
• The sail breadth based on the original traces balances with the hull;
• The original rudder functions with the ship;
• The ship manoeuvres well and is able to tack and beat to windward.

32The challenges involved in test sailing Saga Oseberg have led to a better understanding of how an investigative approach to a full-scale reconstruction and its preserved details can be achieved.

33A full-scale reconstruction of an archaeological ship is an unknown product and the test sailor is, as a consequence, a novice on board. Preconceptions and experience are inevitable but should be used indirectly and only to ask the right questions.

34It is all about how the ship prefers it and not how the test sailor prefers it. The test sailor is a translator of the implicit knowledge the ship holds. The way this specific ship was hand­led originally is lost in time, but can be retrieved by ­studying the details in the original ship’s construction and wear in context.

35All ships have a personality that will reveal itself with use. If the ship does not manoeuver effectively, the test sailor must keep asking questions and try to understand the deeper relationships between the ship’s parts and details. Doing this means that one must be willing to change one’s preconceptions, in order to let reconstructions ‘speak’ for themselves.