Thegns of Mercia
  • About Us
  • Services
    • Living History >
      • Events Calendar
    • Talks & Workshops
    • Reconstructions >
      • Recent Reconstructions
  • TV & Film
  • Blog
  • Contact Us

Shield of an Anglo-Saxon Prince - Part 2:  Building the Shield

19/3/2022

Comments

 
Picture
Aside from the magnificently decorated, heavy shield from Sutton Hoo Mound 1, remains of early Anglo-Saxon shields suggest they were typically relatively plain. Curiously the shields from the other treasure-filled princely burials – Taplow, Broomfield, Prittlewell, Sutton Hoo Mound 17 and others appear especially so, not befitting the status of these burials, with little in the way of decorative fittings, and very minimal, unusually simplified bosses. 
​
In the previous chapter (link) we revealed that (in contrast to the wide variability of shields from contemporary graves) the late 6th century princely burial shields were all practically identical, with suites of four simple disc mounts on the board, simple 1a(i) iron grip reinforcers, and innovative SB-4b / Dickinson’s Type 6 shield bosses – the smallest and lightest of all Anglo-Saxon bosses. In a number of these cases the boards were also made of ultra-light-weight willow. This is the lightest combination of fittings possible, among those evidenced from early Anglo-Saxon graves.   We have therefore argued that the princely shields represent a class of very carefully made, high-performance versions of the standard Anglo-Saxon shield, with weight-reduction prioritised over ostentation.
​
In 2021 we undertook a project to reconstruct such a shield – to explore precisely how light such a shield could be for a given diameter, and to explore methods consistent with archaeological clues which might have been employed to embellish such shields, commensurate with the status of their owners, without compromising their performance.  The result would provide a theoretical minimum weight for an early Anglo-Saxon shield of practical size, and represent our tenth and most ‘authentic’ shield reproduction to date. 


Picture
Step-by-step updates on this project were shared on our social media pages from April to July 2021 organised under the hashtag #ThegnsShield10.

All work on this replica was undertaken from April to July 2021, by Thegns of Merica team members Dr Andrew Thompson and Æd Thompson unless otherwise stated. 

Choice of Wood

Thanks to data from Dickinson and Harke (1992) we had known for some time that willow (or black poplar, which is archaeologically indistinguishable, though structurally inferior) was the most common wood used for shields from early Anglo-Saxon graves, accounting for around 1/3 of wood traces associated with shield fittings. It is the lowest-density of shield woods from early Anglo-Saxon graves (approx. 430kg/m^3) – approx. 75% of the density of maple, and 65% of the density of birch, ash or oak (Dickinson & Harke, 1992). Despite this lower density, it has impressive toughness (following closely behind birch, ash and oak) and moderate split-resistance exceeding that of oak and alder. Anglo-Saxon shield-makers would certainly have had an intuitive understanding of the properties of woods available to them, and certainly would have had a sense that among all available woods, willow had perhaps the best combination of toughness, springiness, and lightness, while also being relatively easy to cleave into narrow planks. The timber of native UK willow species all share relatively similar properties; the most likely candidate species for the willow represented by trace remains on shield fittings from early Anglo-Saxon graves are the often massive-growing white willow (Salix alba), with less likely candidates being the (smaller) goat willow (Salix caprea) or crack willow (Salix euxina and hybrid 'common crack willow' Salix x fragilis). 

​Despite its widespread use and favourable properties, however, to our knowledge willow has only very rarely found its way into modern replica shields due to difficulty in sourcing broad seasoned willow planks.  In the UK today, at least, willow is relatively uncommon, and willow trees of a sufficient size to yield shield planks are rare, much loved as amenity trees, and often subject to tree protection orders (TPOs). When substantial willow timber (usually Salix alba) reaches the market, it is quickly snapped up, particularly by cricket-bat manufacturers. The traditional cricket-bat making industry has also been responsible for the curation and proliferation of a particular cultivar - Salix alba  'Caerulea' / 'cricket-bat willow' - the timber from which is typical of the species, but which grows more quickly and usually with a more reliably straight stem. 

 We were very fortunate to have been able to acquire, as a one-off, a small set of cricket-bat willow (Salix alba var:Caerulea) planks in 2016; just enough to make a single shield. These were approximately 12mm thick, varying up to 72cm long and varying around 12cm wide – far narrower than we have hitherto used for planked shields. These were stored for a number of years, allowing them to stabilise, and waiting for the right project to make use of them.

​

Board Size / Number of Planks

The size of boards of shields from early Anglo-Saxon graves was quite variable (see our earlier article discussing shield sizes - link) but unfortunately there were no clues surviving from which to infer the sizes of the shields from the princely burials. Our willow planks were just long enough to allow for a shield diameter of around 70cm, which is the lower limit of Dickinson’s “large shield” class, and (based on more reliable ‘maximum size’ data) appears to have been roughly the average size of Anglo-Saxon shields (Stephenson, 2002). 

As evidence for the boards of Anglo-Saxon shields themselves are mostly limited to the tiny mineralised remains on the backs of iron fittings, information concerning construction – and particularly, the width and number of planks used for early Anglo-Saxon shields remains elusive. Being very thin and seemingly held together only by edge-to-edge glueing, it has long been supposed that, for optimal strength, shield planks should be as wide, and as few as possible, which seems intuitively obvious. Dickinson and Harke (1992) cite an instance of an early Anglo-Saxon shield from Alveston Manor, Warwickshire, where preserved grain suggested a three-plank construction (with the outer planks, unusually, having a perpendicular grain direction), and at least one shield from the related Swedish Vendel Culture – a large shield from the Valsgarde-8 ship burial (7th century) where the central plank was a massive 52cm wide. 
​
Picture
Extant 'Germanic' late Iron-Age shield (pigment trench shield 23) from the 4-5th century phase of the Nydam ritual deposit (Holst & Neilsen, 2020)

​​In practice, however, building a shield from the fewest, widest possible planks brings a number of significant drawbacks. Firstly, wide planks can only be sourced from the very centre of very large trees (in turn subject to the limitations of the species). These trees are obviously precious, the cleaving of full-width planks from them is extremely labour intensive, and the large surface area of such planks means a high probability of the presence of defects. Extremely wide planks – a full section through the midline of the trunk would greatly limit the uses of the remainder of the trunk – generally wasteful of timber - and (effectively what we’d now call rift-sawn) are also particularly liable to warping due to temperature and humidity changes, which would be undesirable for a shield. 
Most importantly, though, and counterintuitively, the use of fewer, wider planks may not lead to a tougher shield, but actually the opposite. With glued joints being quite rigid, any force of flexion of the board is shared between the planks, with wider planks taking a greater share, with the stress directed to the midline of each plank, and particularly those in the centre of the shield. With early-medieval glues being stronger than the wood they adhere to (see later) it is actually with-grain snapping down the midline of wide planks, especially in the centre of the shield, that is a board’s most likely point of failure, as we have learned to our cost with at least one of our previous reconstructions.

​In fact, better preserved shield-boards from adjacent periods and cultures were very often made of a greater number of narrower planks; the Viking Gokstad ship’s shields were made of 7-9 planks, as were the shields from the Germanic Iron Age Thorsberg bog deposit (3-4th century CE) each being 8-23cm wide (Dickinson & Harke, 1992). To these can now be added data from reassembled ‘pigment trench’ shield planks from the latter phases of the Nydam bog deposit (4-5th century CE) which were 83-104cm in diameter, and comprised of 6-10 (but most commonly seven) planks (Holst & Neilsen, 2020). Plank width was very variable both across, and within shields, but based crudely on dividing diameter by plank number, mean plank width per shield varied from 9-15cm, with a median of 13cm.
​
Picture
Seasoned willow planks for Anglo-Saxon shield, arranged with the approx 70cm diameter board planned out
The construction of shields from fewer, narrower planks allows more efficient use of timber, and means the planks can be quarter-sawn – less likely to warp (Lewis, 2011). More importantly, however, it suggests that shield makers trusted their glues (see later) and were aware that shields were more likely to fail from their planks splitting, than from their joints failing. Interestingly, many of the Nydam shields showed evidence of having failed in precisely this way, with central planks having split in two, with the grain, down the plank’s midline.   
​
Our willow planks – narrower than we are used to – were well within the evidenced range, and varying around 12cm, were close to the average plank-width of the Nydam shields; a 70cm board would make use of all six planks.  

Our planks had originally been only very roughly sawn, and as expected, had warped slightly as their moisture content had stabilised. Thankfully, as they began twice as thick as the board needed to be, the spare thickness gave plenty of leeway to shave them back to approximately flat. The edges were planed, placed tightly together ready for glueing, with the 70cm circle drawn out. ​

Glueing the Boards

PictureAnimal-gluing the Bidford-33 shield replica
Anglo-Saxon and Viking shields were constructed of thin planks glued together edge-to-edge, usually with no other structural reinforcement save for skin-product (leather/hide) layers front and back. Including such layers, based on extant rivet lengths, such shields were 6-9mm (or more rarely, 5-12mm) thick in the centre, so the wood planks could only have been 4-7mm thick. This means that, for a 70cm shield, the contact area between the largest planks could be as little as 25cm2; for other plank joins this contact-area would be even less. That these boards held together at all is a testament to the incredible effectiveness of late antique/ early medieval glues.  Frustratingly no residues of such glues survive in archaeological contexts – whatever glue was used must have been archaeologically fugitive. However, the 12th century craft treatise 'Schedula diversarium atrium / De diversis artibus' ('On Diverse Arts') by Theophilus Presbyter describes recipes for a number of adhesives which were known in the early medieval period (Hawthorne & Smith, 1979). 

For our previous planked shield reconstructions, we have favoured  collagen-based glue  (hide glue / animal glue / pearl glue), a form of which Theophilus describes being made from hides and stag horns. These are probably the most ancient of all glues, made from boiled up animal sinews and skins, to produce an amorphous gel of partly broken collagen chains which are desperate to link back up together under the right conditions. Applied hot, it cools to a stiff gel which fixes joints weakly together before drying of the water component leads to the glue shrinking- pulling the joint tighter (self-clamping action) and solidifying to a glass-like hardness. Its reversibility is helpful during manufacturing, and during repair work, but might be a liability in the context of, say, a shield on a military campaign, with the joints potentially weakened by exposure to moisture. An additional problem is that collagen is actually made more soluble in acidic conditions – such as would be provided by the tannic acids present in part or fully veg-tanned hide, so any leather-faced shield held together with animal glue would have to be very thoroughly treated with oils or waxes to provide some water resistance. Importantly for plausibility, such glue would completely dissolve and wash away over >1000 years exposed to soil acids, and any traces which might remain would be indistinguishable from the collagen that would be present anyway from the skin-product covering. In the context of Anglo-Saxon shield reconstructions, we have found animal glue to be highly effective; an earlier reconstruction of the Bidford-on-Avon 33 shield made of thin ash planks, now decommissioned, cracked down its planks and had to be repaired repeatedly, but interestingly it was always the wood, not the glue joints, which failed. 

An alternative described by Theophilus is so-called “cheese glue” derived from the casein proteins present in milk, which he says was used to fix door panels and altar-boards (edge-to-edge, and covered in hide with the same glue, in a manner clearly analogous to shield-boards) producing bonds that, crucially, could not be loosened with damp or heat.
​

 “The individual pieces for altar and door panels are first carefully fitted together with the shaping tool that is used by coopers and vat-makers. They should be stuck together with cheese glue which is made in this way:
 

Cut soft cheese into small pieces and wash it with hot water in a mortar with a pestle, repeatedly pouring water over it until it comes out clear. Thin the cheese by hand and put it into cold water until it becomes hard. Then it should be rubbed into very small pieces on a smooth wooden board with another piece of wood, and put back into the mortar and pounded carefully with the pestle, and water mixed with quicklime should be added until it becomes as thick as lees. When the panels have been glued together with this glue they stick together so well when they are dry that they cannot be separated by dampness or heat. Afterwards they should be smoothed […]. The panels should be covered with the raw hide of a horse or an ass or a cow which should have been soaked in water. As soon as the hairs are scraped off, a little water should be wrung out and the hide while still damp laid on top of the panels with cheese glue.”    

                                    (Theophilus 'On Divers Arts', 12th century  - Translation from Hawthorne & Smith, 1979)

​It's clear that what is being described closely resembles shield-making, and having not used this method before we were keen to try it on this shield. Knowing a little chemistry, it was possible to rationalise Theophilus’ recipe somewhat. The glue described is essentially just milk protein (casein) which has been separated from milk (curdled – condensed by the addition of an acid), concentrated (by draining off the whey) and then re-dissolved in an alkali, forming an almost saturated protein solution - a thick glossy liquid like PVA, which is made and applied cold and solidifies by the slow drying of the solvent. As the water component of this saturated protein solution evaporates, proteins come out of solution (and begin to crosslink) but so does the alkali, thereby causing the pH of the remaining liquid to fall, thus encouraging more condensation of the proteins. When all of the water is driven off, what is left is a hard protein cement, which is somewhat waterproof.

The waterproofness of this glue appears to be the result of the fact that casein protein is not soluble in acidic, or neutral water. Although the glue is applied as an alkaline solution, the dry alkali powder (base) left behind in the glue when it is dry, is sequestered inside the network of dry protein, thus creating a chicken-and-egg problem. Water can’t dissolve the protein because it’s not alkaline, and the base in the glue can’t dissolve, to make it alkaline again, because it’s trapped behind the protein. A happy coincidence in the case of shields – or Theophilus altar boards – is that any water which soaked into such a board would first dissolve tannic acids from the hide, becoming acidic and all the more unable to dissolve the casein glue.
​
The medieval description of the production of this glue is somewhat overcomplicated. For reasons of controlling purity, we chose to begin with simple unpasteurised milk rather than cheese, which we curdled with a few drops of vinegar. The curds (mostly casein protein) were separated from the whey using cheese-cloth, washed with water (getting rid of the more soluble sugars and proteins from the milk) and rubbed under cold water (to harden the curds so they don’t smear) to produce a fine powder, which was patted dry with tissue, and transferred to a pot.
​
Picture
Replica Anglo-Saxon shield underway; seasoned willow boards being glued with cheese-glue (bottom left) two at a time, then joined. Dry bond between board - top right.
The base Theophilus recommended – quicklime - CaO - is highly caustic, but when added to water (slaked) reacts to produce slaked lime ( Ca(OH)2 ). It is a saturated solution of this less dangerous, quite poorly soluble base (i.e., Limewater) which is the actual solvent for cheese glue. It’s therefore entirely unnecessary to start with dangerous quicklime; far better to source calcium hydroxide and make limewater directly. 

The limited solubility of calcium hydroxide in water means that any solution will quickly become saturated – and reach but not exceed pH 12.4, with any excess calcium hydroxide remaining in suspension of powder or gradually settling. This would allow the medieval craftsman complete, reliable control of the pH of the fresh glue without ever having to measure it, or be aware of the chemistry. pH 12.4 is just alkaline enough to allow the milk protein to dissolve, but not alkaline enough to destroy it. 
​
To make the glue, then, we simply added the clear limewater, drip by drip, to the relatively dry powdered curds, stirring and mashing. This stage is frustrating, because even with the help of the alkali the protein is not keen to dissolve, and does so only on the very surface of each lump. This explains why Theophilus was keen to emphasise the importance of rubbing the curds to powder first, increasing the surface area of the curds for attack by the solvent. Once the curds had completely dissolved, forming a smooth thick glue, we then spread the glue liberally on the plank joins, bonding them in pairs, placed carefully against a wall to allow the joints to hold together under gravity, and allowed the glue to dry over 48 hours. We then whipped up another batch of glue, and joined these pairs together.

​

The Shaping of the Board

Once the glue had thoroughly dried, we began to shape the board. At this point the planks averaged 10mm thick, but required further shaving down of any slight steps / misalignments between the planks to produce a smooth surface. We then began work to thin the board, and plane and sand it down to its ultimate 3D shape. 
​
Contrary to many modern representations, 'Anglo-Saxon' (and, indeed, 'Viking') shields were not of uniform thickness, being thinner on the edge than the centre (Dickinson & Harke, 1992; Bullock et. al., 2011). This is demonstrated by differences in the lengths of rivet shanks, of board and rim fittings, compared to boss rivets of the same shields in graves, and best demonstrated by the shield from Sutton Hoo Mound 1, where the many board fittings and copper-alloy rim demonstrate that the board was flat and of constant thickness for most of its width, thinning in its outer few inches, which were also carved to curve backward creating a very slightly dished effect (Carver, 2005).
Better preserved boards, both from the late Iron Age and Viking Age show that shield boards most commonly had edges thinned from the front, creating a slightly domed surface on the front while keeping the back flat. 
Picture
Preserved 'Viking' shield uncovered in 2008 during excavations at the 10th century ring-fortress Trelleborg near Slagelse, Denmark. The planks were 8-9 mm in the centre, tapering down to 5-6 mm at the edge.
Taking into account the fact that thickness at the edge (as inferred from rim fittings) would also include the two layers of skin product and glue (each contributing perhaps 1-1.5mm thick) a shield with a 4mm edge might only have 1-2mm of wood.
​
It has been suggested that the thinning of the shield’s edge might encourage enemy blades to bide into, and be gripped by the edge of the shield, giving its wielder control of an enemy’s weapon in combat (Warzecha et. al. 2017). However, such thinning also has a considerable impact on the weight and manoeuvrability of the shield, for two reasons. Firstly, the outer region of a circle represents a much larger-than-intuitive portion of a circle’s area, and thus, a shield’s mass. For example, in a 1m circle, the outer 12.5cm ring (ie. covering 25% of its diameter) represents 44% of the circle’s area. Carving this zone to halve its thickness overall would therefore reduce the mass of the shield board (not including hide) by 22%, which is significant. The other reason for such thinning greatly improving shield-handling, is that it redistributes a greater proportion of the remaining mass of the shield closer to the central handle, and pivot point, reducing the force (moment) required to manoeuvre the shield and making it less tiring to use in an active fashion in loose combat.
​

Picture
Replica early Anglo-Saxon shield underway - willow planked board glued together with cheese-glue, cut to shape, and carved to have a dished front and flat back, with edges thinning to 3-4mm, and awaiting addition of skin-product layers front and back.
For our shield we chose to thin the centre of the board to an even 8mm, then tapering the outer zone, on the front, gradually down to 3-4mm at the very edge. Weighing the board before and after, we found that this thinning had reduced the mass of the board by around 20%. At this stage we also cut the hole for the boss.

​

The Handle and Reinforcer

Next came consideration of the handle. Handles of Anglo-Saxon shields were made in-line with the shield planks (rather than fixed onto the back) from wood, with the addition of iron reinforcers of various types which survive with bosses in graves. Wood traces on these iron components give clues about the carpentry of the wooden handles, which also varied. Although sometimes formed from the planks themselves (two smaller holes cut, rather than one, with the bridge between them forming the grip) more often it appears wooden handles were carved from separate pieces & inserted by means of carved lap-joints. In these cases, the grain of the grip-wood runs along the iron reinforcer’s length, perpendicular to the grain (of the board) preserved on the ends of the grip-reinforcer where it joins the board planks (Dickinson and Harke, 1992). 
​
The strongest, and seemingly commonest construction approach was to insert the grip from the board front (Dickinson A1 type), sandwiched between the boss flange on the front, and grip reinforcer on the back. This requires that the back of the grip be flat, while it remains desirable for manoeuvrability/pivoting (Warzecha et. al. 2017) and comfort that the front of the handle be rounded, but also with sloping shoulders of the laps so as not to interfere with the boss flange.


​For this shield, we carefully carved shallow recesses in the front of the already very thin board, somewhat off-centre, then carved a grip of corresponding shape, which when inserted would be flush with the back of the board, allowing for neat installation of the iron reinforcer. The front of the grip was then carefully carved down until the boss could sit neatly over it without interference. Though not our goal, the resulting grip meeting these requirements ended up greatly resembling the grips from Nydam. We then fixed the grip into place with more cheese-glue, and turned our attention to the iron grip reinforcer.
​
Iron grip reinforcers were effectively universal on early Anglo-Saxon shields and came in a number of different types, including rarer long varieties which spanned much of the width of the board. Nevertheless, these, being thin and of relatively soft, non-springy iron, cannot have contributed any rigidity to the board. The overwhelmingly most common type – which became universal in the late 6th to 7th century – was a simple short flat strip of metal, either rectangular or bow-tie shaped, with a rivet in each end (Dickinson & Harke, 1992). This was the type preserved among the remains of our princely burial shields, and we chose to produce an exact replica of the reinforcer from the best documented example – that of the Prittlewell Princely Burial (Blackmore et. al, 2019) which was slightly longer than typical. It was carefully shaped to be slightly narrower in the centre than the organic grip, for comfort, and carefully filed and polished.

Picture
New replica Anglo-Saxon shield's handle assemblage test-fitted. Dickinson & Harke type A1 Anglo-Saxon shield grip jointed and glued into place, with recessed roves and 1a(i) iron grip reinforcer test-fitted.

​There is some ambiguity in whether such reinforcers were held in place by true rivets, or clench-nails; reports usually refer to the former, while reconstruction drawings (Dickinson & Harke, 1992; Stephenson, 2002) typically show the latter, and remains we have examined rarely provide conclusive answers on this, owing to heavy corrosion and disintegration or loss of small components such as shanks and roves. This appears to have been the case for the Prittlewell shield (Blackmore et al. 2019), while the well-studied shields from the Tranmer House cemetery (the 6th century cemetery beside Sutton Hoo) varied, with some grips fixed with clench-nails, and some with true rivets, establishing that both approaches are indeed valid (Bullock et. al., 2011).

​We decided to play it safe and use conventional rivets, carefully recessing rectangular roves in the front face of the grip, so that the peened ends of the rivets holding the grip assemblage together would not interfere with the seating of the boss. These fittings were all test-fitted; at this stage the hand-hole was also carefully bevelled and sanded smooth for comfort.
​

The Rim Holes

Representations of early medieval European round shields often emphasise the rim suggesting they had an additional layer of reinforcement, but direct evidence of this for Anglo-Saxon shields is rare. Metal shield rims are well evidenced from Migration Age bog deposits, and high-status shields from the related Vendel Culture, but the only Anglo-Saxon shield with a full, preserved metal rim so far discovered was the kingly shield from Sutton Hoo Mound 1 (Dickinson & Harke, 1992). It has long been assumed that most Anglo-Saxon shields would have organic edge reinforcement - perhaps of hide strips - as might be evidenced by rare finds of metal edge clips, and it has always been common practice for living historians and reenactors to stitch strips of hide to their shield rims through evenly spaced holes drilled in the rim, as appears to have been the case for the Roman shields from Dura Europos, and the Viking Age shields from Gokstad and Trelleborg. No early Anglo-Saxon shield has ever been found sufficiently well preserved for this element to be directly evidenced – such as by a fragment of the rim of the board bearing holes, but it remains highly plausible, and it’s hard to imagine any other way of satisfyingly finishing the edge of such a shield. 

More recently, invaluable insights into what could be called “Proto-Anglo-Saxon” shields, and by extension, those found in early Anglo-Saxon graves, have been provided by research on finds from the long-used late Iron Age martial sacrificial bog site of Nydam, on the German-Danish border. Holst & Neilsen (2020) provide detailed analysis of recovered fragments of at least 26 shields originally excavated in 1994-5 in loose association with the pine boat deposit which are in most respects uncannily similar to the shields in early Anglo-Saxon graves, with planks dendro-dated to the mid-4th century, but with deposition dated by environmental archaeology to the early 5th.  
25 of the 26 shields analysed from this Nydam phase had closely spaced rim holes. Compression of the wood fibres suggests these holes (which have partially closed-up due to absorption of water in the bog) were made using an awl, and frayed fibres of ‘exit holes’ suggest they were pushed in from the board front. Several holes contained degraded organic material possibly representing the remains of the stitching cord itself, and in three instances a secondary line of holes was present inward of the main row around the rim, perhaps representing the application of a secondary rim strip as a repair. Unlike shields from the earlier deposits/phases within the long-used Nydam Mose depositional site, these shields did not have any metal rims associated with them, and their holes are too variably spaced to have been for the riveting or nailing of metal rims.  “Excavating Nydam'' (Holst & Neilsen, 2020) reports the hole-spacing ranges for each of 23 of these shields, which varied somewhat across each shield, and varied widely from shield to shield, perhaps suggesting the holes were made haphazardly without precise measuring, with shields perhaps made by different craftspeople with different approximate spacing preferences. They most typically had 1-2mm holes spaced 8-12mm apart, running the whole circumference of boards, 5-7mm inward of the very edge.

In the absence of similarly well-preserved boards of shields from early Anglo-Saxon graves, these ‘proto-Anglian’ shields are therefore perhaps the closest we will ever get to filling in the gaps in our knowledge of the construction of early AS shields, including the elusive rim.

Picture
New replica Anglo-Saxon shield in progress; adding evenly spaced holes in the edge of the board for stitching, and for comparison, close-up of remains of the edge of shield-boards from the late Iron-Age Nydam deposit.

​For our new shield, we chose to follow the absolute average spacing seen on these Nydam shields - 10mm apart and 6mm in from the shield edge. The holes were awled from the board front with a square cross-section awl, following the roughly square profile of some of the Nydam holes.
​  
Although folk of Migration Period Northern Europe, and the early Anglo-Saxons could certainly have drilled these holes in the relatively soft board-wood using a bow or pump-drill, it is interesting they chose to pierce them with an awl instead. This might have been for greater convenience, but also, in forcing the fibres of the wood apart, rather than cutting them, and compressing them around the hole, awled holes might have been thought to lead to a more durable shield rim.

​The ” Skin Product” Cover

Traces among mineralised remains on shield bosses and other fittings of shields from Anglo-Saxon burials, and finds of shields from related Northern European cultures from the Iron Age to Viking Age suggest animal hide formed an important part of shields structure, in most cases covering both sides of the thin wooden boards. These covers were likely glued to the boards with the same adhesive used to fix the planks together (most likely hide-glue or cheese-glue) creating a three-layer composite structure. ​ Archaeologists use the vague term “skin product”, as it has for a long time been impossible to distinguish between rawhide, half-tanned, or fully tanned leather in archaeological remains (Cameron, 2000).  *

Addendum:  More recently, intensive analysis of organic remains of a small sample of four Iron Age to Viking Age shields from a number of different contexts (a rectangular shield from Borremose Jutland c350 BCE; round shield from Baunegård g11, Bornholm, c275 CE; the shield from Tira bog, Lavia c875CE; and one from Birka chamber-grave Bj850 c950CE) was able to identify the species and tanning treatment (or lack thereof) of the skin-product components, providing valuable insight into this elusive aspect of shield construction. Warming et. al. (2016) were able to establish that in all but one case (the facing of Bj 850 - reported as deer or sheep, but later clarified as being sheep) the hide used for shields was from cattle, and that in the case of the Borremose and Birka shields the skin-product had indeed been vegetable-tanned (oak-bark tanned in the former case, and birch /larch /pine /spruce /willow in the latter). In contrast the hides used on the Baunegard and Tira shields appeared not to have been tanned, had a stretched parchment-like structure, and the latter appeared to have been treated with oils or fats. Crucially Warming et. al. (2016)'s methods were able to distinguish deliberate tanning in antiquity, from in-situ transformation of preserved skin due to tannins in the soil environment, as, for example, sometimes occurs with human remains (with 'bog mummies' such as Grauballe Man).

​With both tanned and untanned hides evidenced across this sample of shields, which bookend our period, these results unfortunately do not narrow the possibilities for how Anglo-Saxon shields were made, but the methodology demonstrated and validated provides hope that the status of Anglo-Saxon shield leathers might one day be clarified. ​

We're grateful to Rolf F. Warming for bringing this to our attention. 

It’s thought that early Anglo-Saxon tanners generally could not produce fully tanned thick leather, but the thin layers on shields certainly could have been partially or fully tanned (Cameron, 2000). Vegetable-tanned leather, being flexible and easily moulded, is easier to apply to a shield face than rawhide, requires less wetting and thus leads to less board-warping shrinkage, and its smooth surface is more easily waxed and polished to form a water-resistant surface. On the other hand, surface rawhide is harder, potentially more resistant to cutting and impact, and cannot be smoothed onto the board without considerable wetting and moulding, which leads to substantial shrinkage and tension, warping the board but potentially increasing impact resistance (Lewis, 2011). Whichever option is chosen, these layers are, by far, the most expensive parts of any shield reconstruction.

Although we have experimented with rawhide, with mixed results, in recent years we have stuck to veg-tanned leather, mainly for reliability. However, we have always been aware that modern veg-tans are themselves very different from historical leathers, being pale in colour (corrected by application of natural oils, sun-tanning, and dyes) and naturally quite soft. 
​
For this shield we were keen to experiment with something more “authentic”, and turned to “The Woodland Tannery” a new social enterprise based in Fife, Scotland, producing leather from locally sourced deer-hide, pit-tanned over at least 12 months using by-products of management of local forests, including oak, spruce and willow bark, moss and peat. They were able to supply a piece of traditionally tanned red-deer leather for the front of our shield just within our budget, with some natural defects. Having been supplied in a dry state without softening with oils, this leather was extremely stiff and tough, behaving more similarly to high quality rawhide, but also very thin, with little spongy flesh remaining behind the tough outer layer.
​​
Picture
Replica early Anglo-Saxon shield underway: pit-tanned red deer leather hide (bottom left) scraped on the back to remove remaining flesh, wetted, glued and moulded over the front of the board (right).

We cut it generously to shape, to account for any possible shrinkage and to keep our options open for the stitching of the rim, then scraped the reverse - still rough with some connective tissues and hair follicles, until it was relatively smooth across its whole surface. 

This, and the back-leather, were applied with yet more cheese glue, and allowed to dry. The board warped ‘forward’ due to moisture from the glue applied to the back, before returning flat.  On application of the special leather on the front, it was soon clear that this would shrink severely as it dried. The result was an almost ‘shrink-wrapping’ effect on the front (as can be seen by the 'clenching' of the excess leather around and behind the edge of the shield due to shrinkage) spreading extremely neatly and without lumps, and adhering strongly. This did, however, lead to a small amount of forward-warping of the board.
​
Certainly not ideal, this bend is a consequence of applying leather with different properties on each side; if the historic leather had been applied to both sides the stress from shrinkage would have been roughly equal on both sides, resulting in a flatter board.

Rim Stitching

Representations of Anglo-Saxon shields (such as c6-7th pressblech foils like those of the Staffordshire Hoard, and the c8th Repton Stone) suggest they had reinforced rims, and examples such as the Nydam boards already discussed provide ample evidence that stitching through awled holes was an important part of this. However, neither the iconography, grave archaeology, nor Iron Age bog shields provide any clarity as to what precisely was being stitched around the rim. 

It has been common practice to stitch strips of leather or rawhide around reenactment shields, especially those without leather facing, as protection from splitting and splintering. However, as most shields from burials appear to have had skin-product on both sides, it’s equally possible that the rim was formed directly of these, with an excess of either cover moulded around the edge and stitched onto the opposite side. This approach, though considerably more technically challenging, would have the desirable outcome of allowing the shield to be thinner at the edge. 
​
Folding the front around onto the back, however, would not lead to a visibly raised rim like those seen on depictions. Conversely, stitching the back-leather around onto the front would, like a separately applied rim, lead to a flap that could catch blades and direct them to cut the vulnerable stitching. 

​
Picture
Stitching of an Anglo-Saxon shield rim. Top left & mid left - iconography showing Anglo-Saxon shields and emphasising a raised rim - c7th pressblech foil from the Staffordshire Hoard, and the c8th Repton Stone respectively. Bottom left - awled holes in the rim of the replica shield, process of stitching the layers of skin-product (right) and the finished shield rim, as shown from the front (centre)

​For this shield it was suggested we try a different, but also more complicated approach (Matt Bunker, personal comm.) The back leather was folded around onto the front of the board and trimmed just short of the stitch holes, then the front leather was folded round onto the back, and stitched down tightly (with coarse linen cord) pulling the layer beneath tight too. This resulted in a smooth continuous front surface, with a slightly raised rim due to the layer of ‘back-leather’ moulded beneath, while the very edge of the board benefits from two layers of leather. The rim being proud of the stitches which fix it together, with this approach, may offer them some protection from being cut during battle. 

Like the other alternatives this method is conjectural and at least equally consistent with the limited available evidence, but might confer some durability advantages, and produce a satisfyingly neat result. In the hand-holes the front and back covers were also trimmed and joined by stitching, effectively lining the hand holes – a detail implied by remains from the 6th century shield from Tranmer House grave 909 (Bullock et. al., 2011). This was the last stage in manufacturing this board, before addressing decoration and installation of metal fittings.
​
We finally fixed the grip and its iron reinforcer in place, peening the rivets over their roves recessed in the front of the handle, on the front of the board, then turned to the shield boss which would cover it.

The Boss

Iron bosses are the defining feature of early Anglo-Saxon shields, are all complex works of smithcraft, and falling into a number of distinctive types, are also highly dateable (Dickinson & Harke, 1992; Bayliss et. al. 2013). The most striking changes in design occurred from the 6th to the 7th century with the transition from wide, heavy, low forms, often with wide apex buttons and (usually) 5 large disc-headed rivets, to tall “sugarloaf bosses fixed with many small dome-headed rivets (Evison, 1963). These sugarloaf bosses, represented in relatively few late-phase warrior burials, and more abundantly in middle to late period iconography, were far more challenging to produce, and appear to have been adapted with the intention of encouraging weapons to glance rather than hit square-on and bite into the iron. This allowed the bosses to be made thinner, with the result that, despite being much larger than their predecessors, sugarloaf shield bosses were of broadly similar weight (Dickinson & Harke, 1992). Bosses became around 25% lighter from the beginning of the 6th, to the 7th centuries suggesting improvements in smithing, making more economical use of precious iron, but also possibly suggesting shield weight efficiency was becoming a greater priority.
​
As previously discussed, this certainly appears to have been the case for the late 6th century princely burial shields, and their bosses are examples of the very lightest type seen in early Anglo-Saxon archaeology – SB-4b (Holilund-Neilsen; Bayliss et. al. 2013) or Type 6 (Dickinson) weighing only 262g on average. They were the very first bosses to integrate small dome (or ‘knob- ‘) headed rivets on a much narrower flange, rather than the large disc-headed rivets which preceded them, which again greatly improved weight efficiency. This boss type arguably represents a not-so-missing link between the low and heavy bosses of the 5-6th centuries, and the lightly built but tall sugarloaf bosses of the 7th, yet they also occur in some of the latest furnished warrior burials suggesting that this small ultra-efficient boss type continued in use well into the 7th century and possibly beyond. 

Picture
Replica Anglo-Saxon shield - late 6th century - close-up of the lap-jointed and riveted handle from the front prior to installation of the boss (top left) and after installation of the boss, from the back (left) showing 1a(I) grip reinfocer and rivet roves, and from the front (right) showing the replica light-weight SB-4b (Dickinson type 6) iron boss made by Jason Green / Wieland Forge.
Most of the princely burials’ shields had bosses of the very narrowly defined SB4-b2 subtype; again, the boss from the Prittlewell princely burial was the example best documented, although it was very badly preserved. The measurements of this, combined with insights from its sisters from the other burials, helped to inform the production of a faithful replica boss by Jason Green / Wieland Forge – who at this point has probably made more early Anglo-Saxon bosses than anyone since the 7th century. This was the lightest-weight Anglo-Saxon boss Jason had ever produced, only a little heavier than its corroded archaeological counterparts.  We fitted the boss to the board with five small dome-headed rivets with small sub-rectangular hand-made roves – a task made far easier than usual thanks to this boss’ lack of overhanging carination. The rivets were placed roughly – not perfectly – evenly around the boss rim, so as to avoid interference with the grip reinforcer. It is this which likely explains why most historic bosses have slightly unevenly distributed rivets.

​

Other Fittings

Other than iron shield-bosses, the overwhelmingly most common fittings on the front of early Anglo-Saxon shields were plain disc mounts / studs. Remains of these (which typically vary from 20 to 40mm diameter, and more rarely up to 80mm) are associated with a high proportion of shields from 5-7th century Anglo-Saxon graves, and, based on positions in undisturbed contexts (and pressblech foil depictions) appear to have been positioned in groups on opposite sides of the shield, often in pairs. Earlier examples were flat, while later 6th to 7th century sets were typically very slightly dished, presumably to more neatly seat themselves on the front of the shield board without gaps. 
Usually iron (or more rarely, bronze) and sometimes tin-plated, these had smooth undecorated surfaces but were clearly intended to be very visual elements on shields.  Their purpose or significance, however, remains unclear. 
​
The possibility that they were over-large rivets for attachment of straps or other functional elements can be discounted, for they are very rarely found with stout shanks or roves, and although there are some examples with shank-remains or where the flat head of a rivet piercing the centre of the disc has been forged flat with it, in the majority of cases, extant discs are ‘perfect’ and unpierced, and many lack any shank or nail remains on the back, whatsoever, leaving it something of a mystery how they were typically mounted. 

Picture
Early Anglo-Saxon shield-board disc-mounts; remains of the disc mounts from one of the shields from the Taplow Princely Burial (bottom left, C. British Museum); contemporaneous Vendel-culture pressblech foil design showing mounted warrior with shield, with disc mounts emphasised (bottom right} and replica mounts for the new shield - small nails soldered to the underside of the slightly dished discs, for attachment to the board.

​For these examples lacking shank remains, one possibility is that their nails, pins or rivets were attached to the backs of the discs by means of tin or silver soldering. This approach was commonly employed on early Anglo-Saxon copper-alloy items such as brooches and hanging bowls, allowing functional elements (pin attachments, and hanging mounts respectively) to be attached to these expensive objects without disrupting their decoration with rivets. It further appears to have been the favoured approach used to attach decorative disc mounts to shield boss apex buttons. Such soldering is relatively strong and can also be deployed, carefully, between iron items, but where this was done, traces would be difficult, or even impossible to identify on archaeological iron. 

The reason for this is galvanic / bi-metallic corrosion, accelerating the decay of the iron in immediate contact with the solder. In the case of disc mounts, this would lead to both pin and disc quickly becoming dissociated from the solder, leaving little trace of the joint ever having been there on the rusted back of the disc fitting. 

That they were emphasised on representations of shields in contemporary Scandinavian iconography (see above) - even though no other form of shield embellishment is typically shown on these depictions, suggests that they had some significance. It's further noteworthy that the Scandinavian depictions typically show a different number and arrangement of discs than is typically seen among shields from Anglo-Saxon burials, enabling speculation that they served some purpose of communication of rank or identity. This would be very difficult to prove or disprove. 
​
Alternatively, it is plausible that such fittings may have been designed partly to confuse enemy combatants; when positioned unpredictably around the board, and in the absence of other clues, such mounts could help to disguise the orientation the circular shield was being held, and thus which way it might pivot when struck, which would confer advantages to the wielder. 

Most of the sets of remains of shields from the late 6th century princely burials included suites of these disc fittings. The shield from Sutton Hoo Mound 17 had two pairs, the corresponding Taplow shield had at least three, and the more degraded remains of the Prittlewell shield included remains of at least three (Blackmore et. al., 2019), which the excavators inferred represented two matching pairs with one disc non-extant. Using the measurements of the discs from Prittlewell, we added four discs, cut from thin sheet and gently dishing them by panel beating. We carefully attached small sharpened nails to their backs using silver-soldering, ready for them to be fitted to the shield. As a nod to the shield from Taplow, we also added a small polished silver disc to the apical button of the boss, and a woven shoulder-strap with small iron oval-loop buckle, looped around the handle.  

Picture
Replica early Anglo-Saxon shield by Thegns of Mercia members Æd and Andrew Thompson, based on remains from late 6th to 7th century princely burials at Taplow, Broomfield, Prittlewell, Sutton Hoo Mound-17 and others. Shown prior to addition of disc-mounts and painted decoration
Picture
Replica early Anglo-Saxon shield by Thegns of Mercia members Æd and Andrew Thompson, based on remains from late 6th to 7th century princely burials at Taplow, Broomfield, Prittlewell, Sutton Hoo Mound-17 and others. Shown with paired disc fittings in situ, but prior to addition of painted decoration

​Now structurally complete, the shield weighed 2.5 kg, of which around 600g was accounted for by the metal fittings, with the remainder (c 75%) attributable to the 71cm diameter board. This is likely to be very close to the minimum possible weight for a shield of equivalent size consistent with archaeological data, and considerably lighter than most other reconstructions, including all of ours (see Table 1) yet the shield feels surprisingly robust.
Table 1: Comparison of Thegns of Mercia early medieval / 'Anglo-Saxon' and 'Viking' shield reconstructions / replicas
Thegns of Mercia Shield​ Replica
Details
Board
Leather
Est Max. Thickness (mm)
​Diameter (cm)
Area (m^2)
Weight (kg)
Bidford 33
Early/low boss
Ash plank
Front only*
7
77
0.47
3.6
Bidford 182
Early/low boss, bronze mounts
Ash plank
Front only*
12
67
0.35
4.1
Princely Shield
Transitional/light boss
Willow plank
Both sides
8
71
0.40
2.5
Hoard Shield
Tall-straight-cone boss. Jewelled mounts. Bronze rim.
Birch ply *
Both sides
8
77
0.47
4.2
Boars Low, Tissington
Tall sugarloaf boss
Birch ply *
Both sides
9
74
0.43
4.0
Æ.T. Viking
Small Cumwhitton boss, medium ash wood handle
Birch ply *
Both sides
9
74
0.43
3.7
AC Lewis Viking
Simple dome boss, long pine handle
Pine plank
Linen * + rawhide both sides
18*
84
0.55
5.7
 * Note that details provided here are not necessarily an endorsement of the appropriateness / accuracy / authenticity of these approaches; our shield-making approach has evolved over time. Further, not all the shields listed were intended to be ‘fully authentic reconstructions’; for example, sometimes ply is used for greater durability, for shields primarily intended to contextualise / display suites of fittings and/or be frequently handled. At other times, leather on the back has been omitted to allow the planked construction to be seen.   For anyone wishing to produce an ‘authentic’ early medieval shield we would always recommend construction from planks of appropriate wood, with skin product on both sides, and of a total thickness less than 15mm. 
At this stage, then, the result was an ultra-light weight but robust shield representative of the remains found in the princely burials, and built from as-close-as-possible historical materials, glues etc. and consistent with our hypothesis that rather than being merely ordinary shields, these examples were carefully optimised for weight reduction and performance. Nevertheless, the shield at this stage still looked rather plain. As previously discussed, the shield represents the single largest display surface in the typical Anglo-Saxon warrior panoply, which surely would’ve been utilised to great visual effect.  Elsewhere in Anglo-Saxon material culture we see the principle of ‘horror vacui’ – that, generally being time-rich and material-poor, early medieval people tended to decorate every available surface of objects of hard-won materials.  Metal figural appliques are very rare, and notably absent from this class of princely burial shields but, despite a lack of direct archaeological evidence (due to poor preservation) painting has always been a firm possibility, albeit a risky one.

Previously we have avoided painting shields, not because we believe it wasn’t done, but rather, because of a lack of direct evidence to tell us, crucially, how the paints were made and used, and what designs were painted on them. However, in recent years new evidence has come to light, allowing us to integrate painted designs into an early Anglo-Saxon shield reconstruction for the first time – to be discussed in the final chapter. 

References

Bayliss, Alex; Hines, John, et. al Anglo-Saxon graves and grave goods of the 6th and 7th centuries AD: a chronological framework. Routledge, 2013.

Blackmore, Lyn, et al. The Prittlewell Princely Burial: Excavations at Priory Crescent, Southend-on-Sea, Essex, 2003. MOLA (Museum of London Archaeology), 2019.

Bullock, Hayley, Alexandra Baldwin, and Jamie Hood. "Evidence for shield construction from the early Anglo-Saxon cemetery site of Tranmer House, Bromeswell, Suffolk." British Museum technical research bulletin 5 (2011): p-15.

Bunker, Matt. [Personal Communication]. 24/05/2021

Cameron, Esther Anita. Sheaths and scabbards in England AD 400-1100. Archaeopress, 2000.

Carver, Martin. Sutton Hoo: a seventh-century princely burial ground and its context. British Museum Press. 2005.

Dickinson, Tania M, and Härke, Heinrich. Early Anglo-Saxon Shields. Vol. 110. London: Society of Antiquaries of London, 1992.

Evison, Vera. I. Sugar-Loaf Shield Bosses. The Antiquaries Journal 43 (1). Vol 43. 1963

Hawthorne, John G., and Cyril Stanley Smith. On divers arts: the foremost medieval treatise on painting, glassmaking, and metalwork. Courier Corporation, 1979.

Holst, Sandie, and Poul Otto Nielsen. Excavating Nydam-Archaeology, Palaeoecology and Preservation: The National Museum's Research Project 1989-99. Syddansk Universitetsforlag, 2020.

Lewis, Anthony. The construction of a ‘Viking’ Shield. 2011 [Online] (URL=https://docs.google.com/open?id=0BxM95pp3Zv4MYzdkN2FmMTYtNmU2Mi00ZmM5LTgzZTYtYzU1YmQ5YzU2ODQ1) [Accessed 17/03/2022]

Stephenson, Ian P. The Anglo-Saxon Shield. Tempus, 2002.

Warming, Rolf Fabricius, et al. "Shields and hide. On the use of hide in Germanic shields of the Iron Age and Viking Age." Bericht der Römisch-Germanischen Kommission (2016): 155-225.

Warzecha, Roland. Et. al. The Viking Shield; Research, Reconstruction Experimentation. (Youtube Series). 2017 [Online] (URL= https://youtube.com/playlist?list=PL_hVs5MjWFV1IwpWIBswZr1pRq-qcpMIA) [Accessed 17/04/2022]
Comments

    Thegns Blog

    Exploring the history, archaeology and cultures of the "Anglo-Saxon Period" (encompassing the Migration and Viking Ages).

    Archives

    July 2023
    April 2023
    March 2023
    February 2023
    November 2022
    June 2022
    April 2022
    March 2022
    February 2022
    July 2021
    April 2021
    March 2021
    February 2021
    September 2019

    Categories

    All
    Andalus
    Anglo Saxon
    Archaeology
    Architecture
    Arms And Armour
    Art
    Auseklis
    Byzantine Empire
    Christianity
    Coins
    Coronation
    Cosmology
    Experimental Archaeology
    Farming
    Fashion
    Festivals
    Food & Drink
    Frankish Empire
    History
    Kingship
    Knives
    Language
    Leatherwork
    Magic
    Migration Period
    Norman
    Old English
    Opinion
    Pagan
    Philosophy
    Poetry
    Princely Burials
    Prittlewell
    Reenactment
    Religion
    Roman
    Seasons
    Secrets In The Stones Series
    Shields
    Sociology
    Sutton Hoo
    Swords
    Symbology
    Taplow
    Thegns Reconstructions
    Trade
    Traditions
    Viking

    RSS Feed

Powered by Create your own unique website with customizable templates.