Investigation 10
A Personal Account - Paul Constantine
Table 19 - Useful or Useless? & Believe it or not?
When the reconstructed ship is built it is important that nothing has been forgotten or left out of the research side. It is not possible to relate here every question that has been asked, or every problem that has been investigated, but this examination of the rivets illustrates the approach that has been taken. Nothing has been overlooked for the want of trying or for the want of effort; both physical and mental. Nothing has been left to chance.
In Investigation 6a Joe Startin discussed potential sources of information relating to rivets and Table 19 in Volume 1 was one of them. Table 19 cropped up again in Investigation 8 where that Table was illustrated (as below). The Table shows the positions of some rivets and the distances between them. Could this table be used to determine the widths of the strakes at different positions along the ship? There has been a lingering suspicion that Table 19 might possibly hold a hidden key to plank widths from the position of those rivets catalogued in the Table. Joe said that it could useful for ‘showing average widths and general trends’ which is what I also hoped. Eventually, Joe concluded – ‘a glance shows that the information would be of little value’ and also, ‘I assume no useful information regarding hull shape has been drawn from the rivet-plan data.’ More about the ‘rivet plan’ below. To outside observers these might seem to be rather rapid conclusions, but opinions such as Joe’s are based upon lengthy consideration.
I too looked at Table 19 hoping that some useful information might be gained and these were some of the observations about making choices regarding the numbers given in the Table.
This is Table 19 (see Volume 1)
The Table
Rib Numbers (left) start at rib 5. The missing ribs 1 - 4 can be interpreted once the other ribs are completed.
Strakes at top of page. K = Keel is centrally placed with other strakes numbered 1 (garboard) to 9 (gunwale) on either side.
Averages down right-hand side are not really significant at this early stage of adjustment as there are so many missing numbers (e.g. Rib 12 is an average of only 3 numbers, instead of the full 18 possible if the keel is not included).
Totals It is suggested that when the figures have been adjusted it would be more informative to replace right-page ‘Averages’ with the resulting, total-accumulated widths of the Starboard side and, separately on the left side for the Port ones.
It must be stressed that the Table only shows ‘The width of the strakes measured on the line of the rivets’ which is understood to mean the ‘distance from rivet to rivet across the strake’. To find the total strake width up to an inch of width outside the rivet could be allowed. This total board width, the overlap of the lands and back-bevels is being ignored at this stage. Only a single rivet for measurement purposes lying outside row 7 survived to be included in the 1967 Table, so only rivets out to that 7th line will be included in this discussion. The top line on the Table across the craft is at Rib 5.
The dimensions were recorded with a degree of accuracy by excavators. They mostly used a tolerance of ½ an inch. The measurements will not be perfect representations of the ship as it was put into the ground. Soil and rivets may have moved during burial or excavation and by hull distortion. As found, rivets were in different alignments, sometimes damaged, sometimes missing. This means that in some places a value judgement may have been made by the person choosing to which location on the rivet to measure to (see Investigation 6a this website). It may not have been possible to produce totally accurate dimensions by always seeking to measure to a particular point (e.g. head/point) on each rivet. Measurements may have been slightly incorrect.
However, the ‘truth’ of what was measured is evident in the figures, as certain discrepancies which could have been removed, have not been readjusted. An example of this could be in Strake 4, Starboard, where the recorded numbers (reading downwards) are:
9 -7½ - 9 -9½ -10½. The number 7½ is immediately suspect from the point of view of plank width in this sequence and possibly the result of a rivet that has been misplaced. This could be corrected, but it wasn’t, it was left as measured. The table shows large areas where widths are not recorded. This clearly shows the areas damaged in the bottom of the ship caused by people walking about in the impression. The information about the top two planks is almost non-existent. Despite these drawbacks the table might show clear trends in plank dimensions and possibly information about the original craft.
Unusual Pattern?
The Table appears to indicate an unusual pattern of strake widths. Modern wisdom would expect that strakes close to the keel of the ship would be wider than those positioned further outboard. According to the table, this is NOT the case with this ship. The ‘Averages’ at the base of the Table may be averages of very incomplete figures, but it is clear that strakes 1 & 2 are relatively narrow and strakes 6 & 7 are considerably wider. Could this be a fundamental observation relating to Anglo-Saxon ship building as represented in this craft that must be reflected in any reconstruction? Construction is one of the prime purposes of this experimental archaeology undertaking. It is these doubts and suspicions about the table that make it intriguing and worthy of close examination. Is the table useful … or useless?
The information given in the Table could be enhanced by seeking to transpose dimensions from one side of the craft to the other and by looking at sizes that might be modified, such as that mentioned in the Strake 4 above.
Cross-checking
The missing strakes at 8 & 9 offer some freedom for adjustment of strake widths, for there is another way to cross-check for overall cohesion of the figures, as follows:
- By reference to the naval architect’s Table of Offsets the ribs can be drawn
- From the drawn ribs, the length of the curved outer surface of the rib, to which the strakes are secured, can be determined. This might be called the ‘Skin-length’.
- When the combined strake widths are determined they can be compared with this outer rib dimension to see if there is agreement and/or surplus between the figures produced.
Dimension transfer
Whilst approximate sizes can be transferred from one side to another to produce an approximate ‘balance’ of figures to Port and Starboard of the craft, they can also be read vertically to see how they affect the numbers on the ribs above and below them. This will indicate the form of the strake and the smoothness of its shape as it must be in approximate agreement on these adjoining ribs that are, on average only 3ft away. The object of the adjustments is to produce a harmony of figures both across the craft and down each strake that is in acceptable agreement with the ship’s skin-length dimensions as produced from the offset figures.
Manual Method
Begin at the top with Rib 5. Currently strakes 8 & 9 can be ignored as only a single figure exists for them in total.
Rib 5
This is being used as a typical example of some of the choices to be made. It is not desirable to make full written explanations of every decision, but some, such as these can be covered. These are all the numbers in the Table for the first recorded, Rib 5.
PORT STARBOARD
7 6 5 4 3 2 1 K 1 2 3 4 5 6 7
8 7 7 8½ - - - - 6 7½ 8½ 9 8½ 7 -
The first figure that could be moved is the 8” from Port to Starboard on strake 7.
Other figures can be taken from the Starboard side to fill in blanks on the Port side.
Looking at just central numbers line 4 to line 4 it might be possible fill in some blanks with ‘expected’ numbers.
7 6 5 4 3 2 1 K 1 2 3 4 5 6 7
8½ 8 7 6 7 6 7 8½ 8/9
It will be seen that transferring numbers from right to left produces an acceptably smooth series of numbers approximately balanced on either side. Starboard Strake 4 is recorded as 9, but it is the only high figure on the rib and could probably be reduced as figures outside it are lower. As soon as this is done the new figures should be compared with the figures on the next rib down.
Are they relevant?
Figures on each side can also be added up. Is the Port side larger or smaller than the Starboard?
Can small adjustments be made to bring the numbers closer together whilst remaining relevant to all the other figures around them?
The keel measurement is shown as 7, because the Table number below it on Rib 6 is shown as 7. The sequence, reading down the keel is 7 – 7 - 6½ - 7 - 6, so it will make sense to record all these as 7 at the forward end of the ship. However, it is worth noticing by reference to the drawings of the ship, that from Ribs 8/9? forwards, the keel shape is changing from 'flat' when amidships (lower left) as it transitions to 'deeper/narrower' (upper left), to ‘taller/narrower’ as it was modified to meet the stem timber (see diagram right). The point where this tapering begins is a matter of choice. Volume 1 indicates that it is already happening ‘towards’ rib 8 (see above left). At Rib 5 the ‘keel’ is already the stem and it would probably maintain its width at this point all the way forward as it has to be hollowed on the inside to provide adequate fixing for the garboard rivets. This is consistent with the Stem-post detail (right) which shows it as being 7” wide inside at Rib 1.
Whatever the choice of tapering, the keel was thought to be wider in the middle than towards the ends according to these diagrams. The examples given (above left) show the overall width of ‘Towards rib 8’ as being 8½” wide and the midship section (below it) as about 12” (10” between rivets) wide. These dimensions confirm the assumptions being made within a ½” tolerance.
There is some freedom of choice where many dimensions are unavailable due to damage in the central keel area. The indication is that rivet-to-rivet dimension on Rib 13 should not exceed 10”?
The measurements towards the stern of the keel clearly show that the keel itself was slightly narrower with a sequence of 6½ - 6½ - 6 - 6½.
Whatever the keel width, half of it has to be added to the calculation of total strake widths Port & Starboard to arrive at a ‘skin dimension’ length from keel to gunwale.
Suspect measurements
Rib 12 on the Starboard side has a single measurement of 8” on Strake 6. To make such a measurement there must be two rivets in place to measure between. The immediate area around this measurement has no other dimensions, implying that the impression was damaged in some way when the measurement was taken. Other dimensions along this strake 6 are in the region of 10 - 10½ inches and this ‘8’ measurement is located close to the widest part of the ship. It is probably justified in regarding it as an incorrect number. It is misleading, but it can be checked (see below).
Similarly, on Port strake 4 there is a dimension 14” in a strake that is indicated as being 10” possibly 11” wide at this point. As this ‘14’ number is 2-3 inches wider than almost every other number in the table it is, again, probably justified in regarding it as misleading.
Seeing the rivets (Rivet-plan Data)
It is possible to see the 8” measurement that stood in such isolation. The rivet cards in Volume 1 show it. Card 5 (see left) shows rivets 1116 and 1114 in a totally blank white area as defined in 1966/67. Card 1 has the scale. Of great assistance are the dotted lines that indicate perceived edges of the strakes making it possible to see the rivets in Table 19 illustrated and in the light of this, it must be accepted that Table 19 cannot provide totally clear information.
The damage was too great when the measurements were taken and the hopes that we once had for Table 19 are beginning to diminish. But, to keep hope alive, might it furnish some indications of the proportions of strake sizes as used by the Anglo-Saxons?
We know that they did not work from precise drawings as we now do. They worked from rules of thumb and proportions. If adjustments are being made to harmonise groups of figures by increasing or decreasing them, it is possible to look at the originals to gain a greater understanding of why some movements had occurred and in which direction.
A photomontage in the back-cover pocket of Volume 1 of Mercie Lack’s pictures of the Port side, provides a visual image of the ship in its pristine state and the uniform quality of the craftmanship that went into its construction. The ‘Rivet Cards’ dotted lines of the strake edges would not have been wavy had there been time to properly survey the ship and retain the drawings. Just look at the wavy lines of the strake edges towards the bottom of the diagram. They are telling us that what was left when the measurements were taken hardly approximated to the situation on first excavation. The rivets moved with the waves.
At some point an accepted, proportional strake widths can be applied to the lines of the ship, but the outcome would then depend on, which Ship Lines are to be used? We have several to choose from (see last diagram on this page). Whilst plank runs can be laid onto the ship’s Lines it would be very unsafe to determine precise plank runs from the rivet positions illustrated on the 1966/67 cards alone. Rivets can be laid onto plank runs. There is an order here. 1.The Ship Lines. 2. Plank runs. 3. Rivets. To lay out plank runs it would be helpful to have outline proportions reflected by the numbers. They are not precise, they are indicative – rules of thumb and proportions, as the Anglo-Saxons used.
There is additional very important question to bear in mind - How was this projected ship’s plan on the Cards drawn? A true plan i.e. bird’s-eye view, would distort the actual distance between rivets that are positioned on the upward curving sides of the ship. For the measurements to be an accurate representation, the sides of the ship would have to be ‘rolled down and outwards’ on either side to make the image flat, sometimes called a ‘surface development’ or ‘surface net’. It would seem that this latter method must have been used, otherwise the scale would be useless for any transverse measurements.
Learning from Saxon Sudoku
By carrying out the process as described for Rib 5 (above) on each successive rib, it is possible to fill in many ‘expected’ numbers across the ribs. Blank areas still remain. Working up and down strake lines give a number of indications and then, comparing the front of the craft with rear, will furnish others. It is an interesting learning process.
Stern numbers
The initial impression is that numbers are sparse towards the rear of the ship, but ribs from 19 to 22 give a clear set of dimensions to anchor the others around them. Rib 20, in particular has 13 numbers and by transferring to the opposite side there are 16 numbers to be considered on this rib.
Once these numbers have been inserted it is possible to look up towards the other (bow) end of each strake to find out how the numbers there, relate to the stern numbers. There is a progression of numbers that make the visualising of the strake possible. Three strakes on either side of Rib 22 are particularly illustrative and they can influence other strakes around them.
Pairing
Pairing strakes can also produce some interesting figures. If the middle of the ship is at strake 13, then, on a double ended ship, strake 12 and strake 14 should have similarities, as will 11 & 15, 10 & 16 etc. Great care will have to be taken with this approach, especially as Paul Handley identified a difference in shape between the ends, with the forward half being somewhat fuller than the rear. They may be a long distance apart, but Ribs 20 & 6, 21 & 5, can be compared. It is proportions that are being sought, not total mirror images. This leaves Ribs 22, 23, 24 etc. to offer a few numbers that could relate to the missing forward ribs, beyond rib 5. Looking at this last point, Table 19 tends to present a slightly unbalanced view of the ship’s form as the numbers go right to the stern at rib 26, but there are quite a lot missing at the bow ribs 4, 3, 2 & 1.
Totalling numbers Port and Starboard
When sufficient numbers from both sides of the hull have been decided on any rib, each half can then be totalled to see how close the opposite sides are to each other.
Following a first run-through these were some of the rough results for Port and Starboard.
Rib 5 56 ---56½
Rib 6 59 --- 57 Strakes 8 & 9 would still have to be added. They are not the same each side, of course,
Rib 7 53 --- 59½ but they are acceptably close.
Rib 8 67 --- 65½ The worst one is 4½” different, but if the difference is split, they are within a couple of inches
Rib 9 71 --- 73½ of each other and an adjustment could easily be made. The small changes needed can come
Rib 10 73 --- 71½ about naturally, as each strake shape is reviewed from top to bottom of the Table.
Rib 11 74½ - 71½
Girth Measurements
When the totalled dimensions above are compared with the girth at each rib, that is, the curved distance from keel to gunwale they will be seen to be too short, as they only went from line 7 to the keel. So, if we say that Rib 8 has a girth measurement for one half of the hull of (say) 82inches, then we can subtract 66ins (rough average of both halves) from that, to find that planks 8 & 9 (gunwale) should fill 16ins. It is possible that:both planks could have been the same width. One was wider than the other, the width of some of the planks recorded lower down have been either too narrow or too wide. Only when all the figures for all the ribs and all the plank runs have been calculated and all have been harmonised with each other can some kind of pattern be established.
Believe it or not?
If you have managed to follow all this explanation to this point – Congratulations!
It is time to take stock and there are several points to make here. The first concerns the complexity of the investigation.
Anyone coming new to the problem of building this ship might think that it is just a simple matter of taking some known measurements and starting to build. It is not so. When we say that we have been working on it for 6 or 8 years people wonder what we have been doing all that time. The material that you have been reading relates to one tiny corner of the investigation, so you may begin to appreciate the effort required to push the whole project forwards.
Table 19 was published in Volume 1. It cannot be avoided. Does it hold some hidden nuggets of golden information? Or is it of very limited use? To investigate it properly means understanding clearly what it was saying and that means understanding the original recording method for the rivets (see Investigation 6a Measuring). This was not easy as the descriptions of the methods used are sometimes ambiguous and needed close analysis.
Well would you believe it? The closer one comes to a problem, the more one works on it, the more one may lose sight of the overall picture. It is difficult to stand back and think afresh. In engaging in a lengthy investigation of this nature, early, small decisions are made that will affect future outcomes. Layer upon layer of compromises, averaging out, give and take, all obscure the start of the process when the end is approaching. How many guesses have been made? How many estimates? How many averages? By pushing numbers all along the line, it will be possible to arrive at solutions – BUT – can they be believed? How good was the fundamental evidence upon which they were founded?
No stone unturned
If you have been expecting a nice cut-and-dried table of results and a single, simple message you will now be disappointed. For you, there should perhaps have come a time at some point in this process when a reluctant realisation has dawned. The information in Table 19, intriguing as it appears, was gathered at a time when the ship’s impression was too badly damaged to be a reliable guide to the original rivet positions. The rivet cards in Volume 1 confirm this.
Back to Basics
With the basis of Joe Startin’s calculations we can probably say that the original 1939 drawings were made based on hardly more than a couple of hundred approximately (by today's standards) measured rivets. However, these drawings were made by a skilled man A. S. Crosley who had been inside the ship’s impression, had been suspended over it and, had additional notes to assist him in trying to represent the form of the hull that he must have known better than anyone else. He was happy enough to give talks about the accuracy of what he drew, including the damage and imperfections and he answered queries from a discerning audience about it. He had the experience to call upon and today we must reluctantly come to the conclusion that there is no greater degree of accuracy for us, than his. The people involved in hurriedly surveying the ship did a sterling job in the exceptional circumstances that they were working under.
The single most problematic feature of the ship was that one of its ends had separated and had fallen apart. It seemed a simple matter to just to return the sides to the way that they should have been. In so doing, the door was opened for a range of solutions to be suggested, something which continues today. Below are just three versions of one side of the bow, though it was the stern that was damaged (see Investigation 2). On the basis of what you now know, which solution appears the most likely to you? This ‘blind-testing’ choice is one that has to be made and it is not easy or simple. Your guess could be as good as any other.
The final words might be left to Angela Care-Evans who compiled Chapter 5 of Volume 1. Writing in March to Keith Turner who worked for the National Trust at Sutton Hoo and was instrumental in attempting to organise a reconstruction of the ship in 2004, she made her opinion quite clear. Referring to a previous meeting she said:
‘Discussion centred around re-examining the evidence for the 1965-67 British Museum excavation of the ship and the three-dimensional recording of the in situ rivets. This to my mind will not advance our knowledge of the ship’s hull at all as by the time the excavation took place the hull had suffered considerable damage by not being back filled in 1939. Much damage was done by the army and the overall attrition of the upper strakes through weathering and the subsequent movement of the rivets which were loosened by excavation means that the record in 1965-67 was (pace Jan-Bill) of an essentially wrecked hull and that the rivet positions, although recorded in three dimensions are fundamentally flawed.’
Jan Bill is an expert in Viking age archaeology in Denmark & Norway.