A personal account - Paul Constantine
Hull Shape 2 - Measuring
Defining the hull shape of the ship (2)
Joe Startin has done a great deal of research relating to the ship and produced a number of detailed papers. To the uninitiated it would seem a simple matter to decide what shape the ship was and then reconstruct it. This is not so. The story of the ship, its hurried excavation and subsequent damage makes it very difficult to come to a definitive form. If ten experts were to examine it, they would come to at least ten different interpretations. In the final analysis whatever shape is constructed it will be an interpretation. Strictly speaking, even the very first drawings made in 1939, upon which all others should be based, were interpretations. Joe's draft document was dated October 2018,
What follows may give just a small indication of the complexity of the issues to be faced. They would near impossible to understand without Joe’s explanations printed in black below.
Explanations by Paul Constantine are in blue.
Concerning the hull, there is no more data than they had in 1975.
It is known that the people behind Volume One (1975) were immersed in the data, and very capable. We know that the published results in Bruce-Mitford, 1975, did not convince some experts that the information for a useful reconstruction was available.
We believe the value of the 1939 drawings has been underestimated.
We have computer-aided methods to help us handle the available information in ways beyond the scope of the situation in 1975.
As we have no new data, we have to be very clear about our understanding and the use of what we have available. Whether this is claimed to be the position of a location on the outside of the hull, or the position of a certain part of a rivet, it must be explained how we got it and how much reliance we want to place upon it. The explanations may be long and tedious because of adjustments made for individual archaeological features, varying levels of uncertainty and compromises made between different sources.
What is the 1939 data?
Philips 1940a, 1940b and 1940c give useful descriptions and some general idea of overall dimensions. As a guide to hull shape they are very limited.
What is useful is what is available from the ‘Lines plan’ included in Science Museum, 1939.
There are no working papers to back up this ‘Lines plan’. Any associated Science Museum records and any records from Commander R K D Hutchison who led the survey are known to have been destroyed. Hutchison was a retired Lieutenant Commander who was Keeper of the Department of Ship’s Models at the Science Museum. He died serving as an Acting Commander in 1944.
There is a paper, Crosley, 1942, which describes some of the survey process.
A S Crosley was a Chief Draughtsman at the Science Museum. He died in 1981.
(PCon) see Investigation 2, this website - Lt. Cmdr Hutchison & The lost drawing - initials a.s.c.
The British Museum has a large, high-quality photographic record of the 1939 ship. These photographs were mostly taken by the well-equipped and capable amateurs, Mercie Lack and Barbara Wagstaff. Many of the choicest examples can be seen in Bruce-Mitford, 1975 Volume 1.
- 1 Observations on the 1939 ‘lines plan’
My interpretation of the ‘Lines plan’ is that it is to the outside of the hull. This is speculation. My feeling is that Hutchison would be inclined, by temperament, to take a traditional approach. It also has practical benefits, because it leads naturally on to consideration of the displacement and the stability of the hull.
The description of the survey in Crosley, 1942, is sometimes not very clear. From the record:
A.S. Crosley’s Survey of the 6th Century Saxon Burial Ship. Read at the Science Museum April 21 1943. Rove & Rivet
- (PCon comment) Crosley relates how the interior of the ship had already been scooped out by the ‘treasure excavators’ (his words) when he arrived. The rows of ‘nails’ (rivets) were the chief guide in the survey. They were very fragile as the sand would easily trickle away from them and when it rained a number would fall down into the base of the excavation.
A Datum line was stretched the length of the ship 10ft above the keel. Tension was maintained by securing it to stayed, upright posts outside the ship at either end. But there must have been a catenary (sag) in the middle to be allowed for if it was used for vertical measurements. Its function must have been mainly for alignment.
They had two 4ft tall post with heavy wooden bases made. A slot in the top of each upright housed a 14ft long, wooden beam that could slide back and forth within the slots. The uprights were positioned about 5ft apart across? the keel of the ship (Note. In his paper Crosley says the frame was set up across the ship, but all pictures show it being set up along the length of the ship. In practical terms it would make better sense to have it running along the keel. The word ‘across’ could be incorrect?) The rail was added, then its position was checked against the overhead Datum line by using plumb bobs. (Plural in his account, further reinforcing the surmise that it was along the keel.) When the location and alignment was correct, the frame was secured by using additional guy ropes as stays.
They started amidships and worked towards either end using various, smaller, wooden battens to measure across the ship at the different plank heights between, but close to, each successive rib and the sliding rail.
All survey apparatus pictures
Courtesy of British Museum
- (PCon comment) It is obvious that those involved with the measurements walked about inside the fragile ship impression, which could never happen today, but the extraction of the grave goods had already taken place with the resulting damage and time was pressing, so these methods, which might be considered brutal now, were used. It took three people to measure check and tabulated the results.
To take a measurement one end of a batten would be placed immediately ‘adjacent’ (see later discussion below) to a rivet near to a rib. The other end of the batten was lifted or lowered to make it level and the length to a plumb line, slung from the sliding rail was noted. At the same time the height from the keel at this point was noted.
The ribs were roughly 3ft apart, so it would have been possible to gather measurements from 3 or 4 ribs, for each setting up of the frame. Each rivet was positioned horizontally and vertically, using a rivet on either side of each rib, on each of the ascending, eight plank rows. There are 26 ribs. (See later calculations)
Where the line of the gunwale could be seen they also recorded that. They used the sliding-rail frame to measure the centre-line distance from rib to rib by using plumb bobs and by sliding the rail along they could check 4 frames from one, fixed position. At the narrowing ends of the ship they rigged up a device that looked like a child’s swing to support one person, to gain access, whilst avoiding further damage.
The description says that they were able to identify a 5° heel to port? (The heel was actually to starboard and he implies this later, so this is an error) and also the distortions of the keel. By scraping the sand away in the vicinity of each rib where it crossed the keel they found keel spikes which they were able to use to define the run of the keel and its relationship to stem and stern timbers.
The ribs were recorded as being 3 – 5ins wide and about 3 to 4ins in depth. Crosley says that the strakes were about 15ins wide. He guessed at possible construction methods for securing the gunwale, but it is clear that he was being influenced by the Nydam ship and other craft as he quotes several sources with which he seemed familiar. There are detailed dimensioned drawings of any metal fittings.
In later questioning, Crosley said that he was responsible for the drawings, which were based on measurements made from the port side of the craft. He thanked the director of the Science Museum for allowing him to undertake the survey and his colleagues Gillman and Jacobs for their assistance.
- Crosley says ‘… a few hundred dimensions were taken on the port and starboard side between the 26 ribs of the ship…’ For two sides, with 8 rows of rivets per side, 26 ribs, and both the horizontal and vertical dimensions, this leads to potentially 2 x 8 x 26 x 2 = 832 measurements on 416 rivets. If allowance is made for the big patch under the burial chamber where the rivets were completely absent, some strakes not continuing all the way up to ribs at the bow and stern and other missing or damaged areas, this does seem consistent. He measured ‘immediately adjacent to a clinch nail near a rib…’ So, where was ‘adjacent?
- Speculating again – by the point of the rivet, simply because it was easiest. Pictures show three men working together in what Crosley calls at the end of the paper ‘difficult conditions’
- If we assume he was asked to produce a plan to the outside of the ship, Crosley should then have adjusted the two measurements of each rivet. (Given the abiding uncertainties, this may seem pointless. However, you would otherwise have to accept, for example, a possible error of around 4 inches in the beam of the boat.) Considering this further:
- The point of the rivet will not be at quite the same position as where the rove sat. The land where the head of the rivet sat will be on the outside of the hull, 2 inches from where the rove sat. Using these outside lands is a slight overestimate of the outside position if you are considering the displacement of the boat, because of the clinker. It would be reasonable to estimate the position of the outside of the boat as 2 inches from the point of the rivet, taking the direction of the rivet.
- (PCon Explanation) Are we nit-picking?
- This is where Crosley's description is ambiguous and subject to debate. The wood of the hull had been crushed or leached away to a thin, black line in the sand, but it still preserved the grain pattern of the timber of the inside of the hull. This black line would probably have been crushed outwards to the dimension of the outside of the hull. Where did they measure to? The inside of the rove projecting above the sand on the end of the rivet, or to the black line underneath the head? Or somewhere in between? Does it matter? Were they being this accurate? Even 27 years later they only measured to a tolerance of ½ inch. The sketch-diagrams to the left attempt to illustrate the problem. The curved ship's-hull impression was supported by the soil. The rivets were projecting from the soil. If they were cleaned right back only the rivet head would be in the soil and free to crumble away. It would be displaced and fall into the impression. A small cone of supporting soil could be left to hold the rivet in place. If we look at 3 rivets in positions A,B,C that represent different rivet positions, we see that they are held at different angles. Now comes the question? To measure to the INSIDE of the original hull, the batten end would have to be held steady next to the inside of the rove. (Red arrows on each of the diagrams). To measure to the OUTSIDE of the hull the end of the batten would have to be placed on the black line of the skin alongside the supporting cone of soil. (Green arrows on each of the diagrams). Why is this important? Because it could add up to a significant difference in the displacement of the ship - that is, its weight-carrying capacity. Looking at the 3 rivet positions it can be seen that if the rivet is high on the side of the ship (A), it might make a difference of about 2 inches (each side). If it is a rivet on the bottom of the ship (C) the measured distance from the centre of the keel would be virtually the same. Well - are we nit-picking? Maybe? We have the luxury of not working against time, not being in a hole in 'difficult conditons', possibly sitting on a 'swing' trying to hold a stick straight without actually touching a friable, loose rivet, recording hundreds of measurements to swinging plumb bobs etc. Put yourself in this position. Think of the way they walked about in the impression. Look at the pictures above. How would you have done it? Can we expect to attach current archaeological good-practice to their methods? Can we say that from the single word 'adjacent' that all the measurements were taken to the inside or the outside of the hull? Joe is an able mathematician and so he suggest a possible mathematical calculation. The angle of each rivet to the vertical is therefore required information. (For an undisturbed rivet, this angle will lie in the plane of the rib.) Such an adjustment for potentially 416 rivets sounds arduous. It need not be that difficult, if you class each rivet as a 0 degree,10 degree, 20 degree, 30 degree rivet etc, by eye, as you go. If you were measuring to one eighth of an inch, the (x, y) adjustments for the starboard side looking from the stern, using simple trigonometry, would then be respectively (0, -2), (+⅛, -2), (+⅝, -1 ⅞), (+1, -1 ¾) etc. and can be applied systematically to all the rivets in each class. The results would be slightly ragged in parts because of the granularity in rivet angle, but good enough. (PCon explanation) Taking Lines Traditionally and even today, when the Lines (the shape of a craft) are to be taken from an existing hull it is standard practice for the shipwright concerned to examine the hull, then choose what is considered to be, (using experience) the ‘best’ side. The shape of only the chosen half of the hull will then be recorded. The reasoning for this is that nobody would knowingly construct the two sides differently, they should always be the same. However, shipwrights accept the reality that small changes may have crept in when a craft was constructed. To avoid conflicting information, they measure just the one side and then endeavour to make any copy have both sides the same again. The recordings used for the drawings of only one side of the craft in 1939 may not have been a total disregard for accuracy, forced by a shortage of time available. Both sides were distorted. Notes regarding additional details such as unusual rivet runs or the location of scarfs should have been noted and then added in their appropriate positions to any drawings of a craft that had two sides to be built the same.
Strake LinesThe drawing gives strake-join lines for both sides, but specific rivets positions only for the port side of the ship. Volume I, mentions this on p. 234-235, and it is evident from a glance at fig. 280, p. 365. Arguing as previously described, measurement information was available for a potential maximum of 208 port side rivets. Other rivet positions on the port side must have been filled in by checking sketches and notes made at the time, (PCon That no longer exist) maybe by using photos and by interpolation.
A glance at the lower strake lines on the body plan shows these are somewhat implausible. The bow and stern do not seem to join up and the width of the garboard strake is unclear. The intended meaning must be drawn from the view from above, the half-breadth plan.
The ship was found to be tilted towards starboard, and the angle of heel varied slightly along its length. It would not be difficult to remove the tilt from any plot that might be made of the measurements along a rib. They would probably have used a drawing office technique.
Information from 1965-70, as interpreted in Bruce-Mitford, 1975
After the war there are two sources of information about rivet locations as they were placed when work began.
Table 19. The 1967 strake widths table
This is information from measurements taken in 1967. It is tabulated in Volume 1 (1975, p. 354). Table 19 is captioned ‘Showing (in inches) the width of the strakes measured on the line of the ribs’. Measurements are actually given to the nearest half-inch.
The table gives a good indication of the state of the ship as found by the 1965-70 excavation. There is no information at all for ribs 1-4 and 26 where the impression was irrevocably damaged. The central burial chamber area is totally denuded. The port side is somewhat better endowed than the starboard side. Crosley, 1943, identifies a potential maximum of 416 measured, rivet positions. This table implies only 201 actual positions remained in 1967. (PCon) Plank widths are measured using only 152 rivets in Table 19.
For the bottom of the ship, where it is comparatively flat, the rivet plan cards (see below) should reflect this data. In fact, a glance at the cards show how difficult these measurements must have been. The variation in the data across the table is explained. Table 19 is useful, but mainly as a way of showing average widths and general trends.
(PCon explanation) The ship’s impression was massively damaged during the war. The top two strakes were totally lost and it is clear from Table 19 and the 1975 rivet plan cards, drawn almost another 10 years later and described below that it would have been impossible to draw up any meaningful plan of the ship from any ‘new’ data. Excavations could possibly re-examine certain details below planks 6 -7, but above that, after 30 – 35 years, the ship was virtually non-existent.
The 1975 rivet plan
The rivet plan is given on the 8 cards in the back pocket of Volume 1. These are plans which show the positions and characteristics (complete, head and shank only, etc.) of all the rivets found in situ by the 1965-75 excavations.
They include other information, such as the positions of identifiable bits of ribs and keel. The implied lines of the strake rivets are given as dashed lines. The datum and the grid used for the measurements they took, together with a scale, are also given.
The lines of the strake rivets are distinctly wavy. Even though the grid could be used to give an accurate position, in plan, of every rivet identified, a glance shows that the information would be of little value.
There is no information given about the vertical positions the rivets.
I assume no useful information regarding hull shape has been drawn from the rivet plan data.
(PCon) The 1939 information was the only source of a Lines Plan drawn by A.S.Crosley, based upon his (by calculation) 208 port side rivets, but also all the other notes that he undoubtably took and his experience of being in the impression of the craft,
There remained the problem of the open stern of the ship that he had accurately recorded.
(PCon) It was agreed by all experts that the ship had been constructed with both ends of similar pointed construction. A well-respected naval architect, Colin Mudie was commissioned to make drawings showing the ship with the sides pulled back in, to illustrate the stern undamaged. The two drawings are described below. Although they are positioned next to each other they are of different scales, which means that they are not easy to cross-reference precisely.
The 1975 lines plan and archaeological reconstruction
These drawings are next to each other as fold-out diagrams in Volume 1.
Fig. 324 is to a slightly larger scale than fig. 325. It shows 25 evenly-spaced sections at intervals of just over 3 feet. Fig. 325 marks off the mid-points of the bottoms of the 26 ribs, and the variations in rib spacing are quite clear.
Fig. 324 shows a dashed line on the elevation which at first sight looks to correspond to the bottom of the ship if the hull was continued around its girth, with no keel, or projections at the stem and stern. When the section lines on the body plan are measured vertically from the sheer line and compared with the elevation, they all bottom out at this line. The body plan on fig. 325 also shows both the plank runs and the section lines and it looks as if the section lines are actually to the inside of the planking.
The plan view in fig. 325 shows the rivet positions for the starboard side as well as the port side. These could have been derived from the 1939 photographs, corroborated where possible with rivets found in 1966-67 (described in Table 19 above) and interpolated where photos showed the rivets were missing. The elevation in fig. 325 show the starboard rivets, whereas the 1939 elevation shows the port rivets.
The fig. 325 elevation shows eight rows of strake rivets, as one would expect. Forward of rib 8 and aft of rib 19, rivets are also shown for the join of the garboard, strake 1, to the stem, keel and stern. (PCon) Table 19 shows that only 5 of these rivets existed at the bow and 5 at the stern, not the number that are present in the Mudie drawing) in 1966/67 This must be intended to show where the internal ‘wings’ of the stem, the keel and the stern project upwards sufficiently for the rivets to be distinguishably away from the groove in these parts. The line above these particular rivets seems to correspond to the dashed line in fig. 324.
The rivets are shown in fig. 325 as diamonds, used in the rivet plans to indicate roves. The consistency with fig. 324 suggests that the lines are drawn to the internal shape and that the intention of fig. 325 is to indicate actual rove positions.
So, fig. 325 does look to be an attractive place to pick off the positions of the internal ends for the rivets. (PCon comment) Most of these rivets have probably been added by Mudie) The considerable effort expended in 1965-70 possibly looking at photographs could then be exploited. The caveats are:
- The rivet positions are marked on a hull shape as dimensioned and faired by Colin Mudie in 1973.
- Mudie’s assumptions about distortions in the hull and plank-spreading are not known.
- Mudie is known to have been dissatisfied with the 1973 lines, possibly submitted to meet a British Museum publishing deadline. He continued working and slightly different lines from 1974 do exist.
- Dimensions such as the distance between lines of strake rivets have to come from photographs, or from the 1939 lines plan. (The evidence available in the ground was poor quality. See sections 3.1 and 3.3)
Note that the positions of the scarfs shown in fig. 325 is a totally separate discussion.
Broadly there are two ways of abstracting data from the record.
The first is the 1939 information, which is a reasonable indication of what was found in the ground, probably to its outer dimensions. There is a lines plan, which is a bit rough as it was produced in a hurry. There is no hard information from working papers to back it up. There is no reason to think adjustments were made for plank spreading, for example. Its credibility largely depends on believing that the survey team were both competent and diligent.
The second is the information from 1975. This mainly benefits from a detailed examination of the 1939 photographs – any additional measurements were taken from a much-degraded object. The nature of the assumptions made to produce the faired shape of the hull are not clear.
Any attempt to reconstruct the hull will be based on data with a number of limitations. Different offerings of hull shape will use and weight that data in different ways. This paper tries to scope the limitations of the data in a helpful way. It should then be clearer how to characterise the input data to a model in a way that allows the outside world to judge where it stands.
BRUCE-MITFORD, R., 1975. The Sutton Hoo Ship Burial, Volume I. London. British Museum.
CROSLEY, A.S. 1942. Survey of the 6th Century Saxon Burial Ship, Transactions of the Newcomen Society, 23(1),109-116
PHILLIPS, C. W., 1940a. The Excavation of the Sutton Hoo Ship Burial, Antiquaries Journal, XX (2), 149-202
PHILLIPS, C. W., 1940b. The Excavation of the Sutton Hoo Ship Burial, The Mariner’s Mirror, 26 (4), 345-355
PHILLIPS, C. W., 1940c. The Excavation of the Sutton Hoo Ship-burial, Antiquity, XIV, 6-27
SCIENCE MUSEUM, 1939. Drawing No 2012/B, Provisional Drawing, 2 Sheets, Sheet 2. Colchester and Ipswich Museum.