Arms & Armour , Vol. 4, No. 1, 2007
A report of the findings of the Defence Academy warbow trials Part 1 Summer 2005 Paul Bourke* and David Whetham**
s e i r u *Cranfield University / ** King’s College, London o m r A e h t f In 1992, Peter Jones established a scientific benchmark for the discussion about the o s effectiveness of the medieval longbow. Since then it has often been employed as the basis e e for those seeking to demonstrate, compare or contrast or re-evaluate the historical role t s played by this weapon system. While the authors of this paper acknowledge the u r Timportance of Jones’s tests in establishing a foundation for the scientific analysis of e h the effectiveness of the medieval longbow, it must also be acknowledged that some of the T ) assumptions in the tests made by Jones are now considered flawed or have otherwise been c (
called into question by shifts and developments in historical opinion. The aim of these
g n tests was to complete a series of trials repeating the work done by Jones to a standard i h s i l that is satisfactory to traditional archery experts, historians, blacksmiths and academics b alike, allowing a new evaluation of the power and effectiveness of the longbow and its u P performance against armoured targets concurrent with current historical opinions from a y e range of disciplines. Once the tests were completed, the team would try and recreate the n a results in the laboratory to provide a basis for future testing. M y b d e h Introduction s i l b u There has been much debate over many years as to the effectiveness of the late P
medieval (14th–15th century) longbow. In addition, there has been some well intentioned but flawed research published by scientists over the past fifteen years with little or no grasp of archery or history, or worse, instances of ‘TV science’ where the requirements of televisual narrative appear to have been allowed to dictate what is recorded as the outcome of experiments (for example, Granada 2003). One of the more respected bodies of work on the effectiveness of the medieval warbow was a paper published by Peter Jones in 1992. It has since often been quoted by those seeking to substantiate their arguments with scientific evidence and has therefore proven to be an influential piece of work in guiding the debate © 2007 The Trustees of the Armouries
DOI: 10.1179/174962607X177436
54
Bourke and Whetham
surrounding the weapon (e.g. DeVries 1997: 454–470). While the authors of this experiment acknowledge the importance of Jones’ tests as establishing a foundation for the scientific analysis of the effectiveness of the medieval longbow, it must also now be acknowledged that some of the assumptions in the tests made by Jones are flawed or have otherwise been called into question by shifts and developments in historical opinion. The aim of these trials was to take the Jones tests as a starting point and to make a new evaluation of the power and effectiveness of the longbow and its performance against armoured targets concurrent with current historical opinions from a range of academic, historians, traditional archery experts and blacksmiths. s eOnce the tests were completed, the team would try and recreate the results in i r uthe laboratory to allow more thorough investigation of impact velocities so o m r effectiveness at various ranges could be determined in a consistent manner as it is Aextremely difficult and inefficient to conduct such tests at longer ranges in the e h t field.
f o s eReview and critique of Jones 1992 e t s u r TFor his 1992 paper, Jones conducted tests using replica medieval arrows loosed efrom a longbow into various unsupported unsupported iron plates at a range of 10 m. This h Twork, whilst thorough and containing valuable analysis of the metallurgy of some ) c period arrowheads, is now seen to fall short in a number of areas and can no ( glonger be accepted as representative of the effectiveness of the medieval warbow. n i hThese areas include the style and power of the bow employed, the weight of the s i l barrow, the choice of arrowhead and the lack of support of the target. u P Based on current historical opinion it is felt that Jones used far too light a draw y eweight of bow, with a modest drawlength — a design similar to Victorian n atarget archery bow rather than a medieval warbow. The Victorian target bow is Mthe pre-eminent ‘longbow’ of modern times. It has a stiff centre section and limbs y bwhich bend. The bow is drawn to the chin (approx 28 in). This makes for a d eforgiving, accurate bow which is not fatiguing to shoot. However, medieval h s i l design bows are said to ‘come compass’. A compass bow is one which flexes as b uone from tip to tip, thereby storing more energy as the entire bow is being loaded, P
including the area of the hand grip. This flex through the grip makes these bows more challenging and fatiguing to shoot and is the reason why they are no longer commonly found, particularly at heavier draw weights, in modern ‘traditional archery’. As a ‘warbow’, these bows are also drawn to the side of the face, usually in excess of 30q . The compass and target longbows are illustrated in figures figures 1a and b. As a result of the chosen bow, the arrows used by Jones were rather small when compared, for example, with those found on Henry VIII’s Tudor battleship the Mary Rose which had an average average shaft length of 30 in (76.2 (76.2 cm). Jones’ experiexperimental arrows were also roughly 2/3 of the weight of these arrows (Mary Rose 2006). It is now widely believed that the Mary Rose arrows are very similar to their medieval predecessors. Jones also used only a single type of arrowhead — a
55
Defence Academy warbow trials
s e i r u o m r A e h t f o s e e t s u r T e h T ) c (
g Figure 1 (a) Compass bow and (b) target bow (Greenland 2001: 13–14). n i h s i l b u Plong bodkin known as a Type 7. However, this head was of a very specific type y e suitable for penetration of only mail or soft armour and which would have been n a practically obsolete by the late 14th century. Jones used this head against some M y quite thick plate armour and, unsurprisingly, found that the level of penetration b d achieved was small. The size and weight of these heavier arrows indicate that e h they not only required a very powerful bow to loose them but also that they were s i l capable of delivering a considerable amount of energy to a target (see Table 1). b u P
Table 1
Impact energies
Weapon
Kinetic energy, J
Sword/axe* 1200 lb 15th century Genoese crossbow2 Longbow arrow* Head 1 (Long bodkin) Head 2 (Short bodkin) Head 3 (Lozenge)
60–130 100 80 75** 86** 92**
*Williams 2003: 945 **to nearest Joule
Bourke and Whetham
56
The experimental approach used by Jones for target mounting is believed to be less than ideal. It was stated (correctly) that flesh on its own puts up relatively little resistance to penetration. From this, it was decided that the presence of flesh behind the target was also irrelevant. Unfortunately, this ignores the effect of flesh in supporting the target. A target plate supported on a flesh / body simulant will behave differently to an unsupported plate. An analogy could be made with trying to press a pencil through a loose sheet of paper — because the paper will move, it can be difficult to penetrate the paper. When the paper is backed with something like a soft eraser, it becomes far easier to push the pencil through the paper as there is now some resistance against which to push even though the eraser s itself provides very little resistance to a sharp pencil on its own. It is felt e i r that supporting an armour plate in a similar way to if it was worn on a body is u oimportant to understanding the performance of the armour against arrow m r Apenetration. In the science of modern body armour testing it has been established ethat a backing makes a significant difference against ballistic and stabbing h t performance (Croft 2003). As a result all police body armour in the UK is tested f o on a flesh simulating backing because of its effect. s e Jones also conducted an analysis of the thickness of period armour from the e t s late 1300s to the late 1400s of German and Italian provenance (Jones 1991: 115). u r TThis analysis covered several helmets (bascinets) but only a single breastplate. e Jones’ helmet analysis sits well with the opinions of one of the UK’s most h Texperienced armoursmiths, Roy King, who corroborates the findings of an extra ) c thick portion at the back of some helmets, indicating that this region is a function ( gof manufacture due to it being the point at which the helmet is drawn out n i hfrom (King 2005). In his magisterial work on metallurgy, Williams conducted a s i l comprehensive study of the thickness of later breastplates (front) for both b uhorseman and infantry applications (Williams 2003). The breastplates surveyed P y were taken from a range of collections from Germany, Italy and the UK to give e na spectrum of European armour from the late 1400s through to the late 1600s. a MWhilst the later data is not relevant to medieval armour it is included for interest y b(especially regarding the thickness of certain pieces). Research conducted by dWilliams indicates that breastplates from the period of the later part of the e hHundred Years War and the Wars of the Roses were around 2 mm in thickness s i l (Williams 2003: 913–915). Other parts of the armour would, of course, have b u Pbeen different thicknesses; for example Table 2 details artefact A.22 from the Table 2
Thickness of Wallace collection artefact A.22
Location on artefact
Thickness, mm
Breastplate Backplate Helmet skull Legs Shoulders Cuisses (thigh) Tassets (upper thigh) Collar
1.3 1.0 1.5 0.8 1.1 0.7 0.8 1.1
Defence Academy warbow trials
57
Wallace Collection (Jones 1995). This information is not included to represent some kind of ‘standard’ but to give a very general idea of the distribution of metal one might find within the same suit of armour. Jones does not comment on the provenance of the armours from which he draws his hardness data. Assuming that this is from the same collection of armour for which he has thickness data (German and Italian), this leads to an interesting comparison with research done by Williams on pre-1400 German and Italian armours (Williams 2003: 62–65, 331–332). This research has found armours of somewhat higher hardness, between 130 and 399 Hv, compared with those armours tested by Jones of between 100 and 250 Hv for the period 1400–1550. s e One of Williams’ findings of additional interest is that English armour from the i r u o mid to late 1300s is, in general, quite soft, from 108–200 Hv, with the occurrence m r of the occasional exceptionally hard piece of 290 Hv and 430 Hv (Williams 2003: A e 731). By the early 1500s, armour appears to be somewhat uniformly harder: h t Williams lists some of Henry VIII’s early armours as ranging from 217 Hv at the f o bottom end to 295 Hv (Williams 2003: 733–735). However, this variety in s e armour hardness is indicative of the problems with research in this area. Most of e t s the surviving armour available to test are unique pieces in their own right and, as u r Tsuch, it is hard to define an ‘average’ armour from any given period. It should e also be noted that it is likely that more of the ‘good’ armour has survived as it h Twould have been looked after even after it had outlived its usefulness. ) c This project concentrates on plate armour rather than mail for several reasons. ( g The limitations of mail armours were beginning to become apparent by the 13th n i h century (Williams 2003: 42, 942–943). At the same time, heavy longbows s i l b were beginning to make their mark on the European battlefield. Gerald of Wales u Pdescribes how dangerous facing the longbow was becoming even before the 12th y e century. At the siege of Abergavenny in 1182 Gerald famously comments on n a how arrows were piercing an oak door 4 in (10 cm) thick. In another incident he Mcomments how a knight was pinned to his horse by an arrow which went though y b his long mail shirt, through to pierce his mail breeches, his thigh, through d e the wooden saddle and on into the horse (Gerald of Wales 1978: 112). Whilst h s i l this sounds similar to the JFK ‘Magic Bullet Theory’ and no doubt has been b u embellished with time, the story still gives an idea as to how little protection P against arrows mail could give. Curry attributes three of the four factors leading to an increased use of full plate armour in the 15th century to missile weapons, citing the growing use of the longbow by the English, the crossbow by the French, and the trend towards dismounted combat, itself a result of the vulnerability of horses to missile weapons (Curry 2001: 422–426). Williams comments on the number of links required to make a knee length mail shirt (28–50,000) and the fact that up to 100 days of labour could be required to make a single shirt (Williams 2003: 43). After the Black Death, labour costs and thus the cost of a coat of mail increased significantly. It appears to the authors that it is valid to believe that a point was reached at which plate armour not only offered better protection than mail but was also cheaper. As such (and as the experimental testing of mail requires much effort and expense in procuring suitable test
58
Bourke and Whetham
material of the appropriate pattern), Part 1 of this study will look at plate armour only. Jones concentrates his investigation on the long bodkin (British Museum category ‘type 7’) style of arrowhead, this head was in use at the beginning of the Hundred Years War but rapidly became obsolete as plate armour became more common because it provided good protection against this ‘delicate’ arrowhead.1 The needle bodkin was therefore rapidly superseded by a shorter, stouter (‘type 10’) bodkin and more bespoke lozenge–shaped plate piercing arrowheads as the Hundred Years War progressed. As in any arms race, as soon as a weapon forces the development of an armour to protect against it, a new weapon to defeat the s new armour will be born out of necessity. There is, however, evidence of these e i r usupposedly obsolete long bodkin arrowheads being in use far longer than the o developments in armour might suggest. In addition to this, bodkin arrows of a m r Adecidedly unfeasible length also exist, for example one is a full 14 cm in length, esurely too fragile to achieve penetration through plate armour of any thickness h t f (Jones 1992: 113). Heads of this length will fail due to buckling and this is almost o s a certainty with an oblique impact inducing a bending moment in the slender e ehead. A square and true impact will also fail the arrowhead by buckling if the t s plate is sufficiently strong enough to resist initial penetration. This continued u r Toccurrence of long bodkin arrowheads which are inferior to, require more e hmaterial and take longer to manufacture than short ‘Type 10’ bodkins, may be T ) explained by work and exhaustive testing carried out by Stretton who makes c ( the fascinating suggestion that this type of arrowhead was used as the core for g ncreating very effective fire arrows (Stretton 2005b: 16–20). i h Arrowheads themselves are exceptionally difficult to quantify in terms of qual s i l bity relative to one another, not only due to the wide ranging design sub-groups u Pbut also the variance within each sub-group. What investigation has been done y ehas often concentrated on hardness and microstructure. This is a fair way of n agrading the material quality as it has been known from an early time that Mthe hardness of an arrowhead is important for its performance. In 1405/6 there y bwas an Act of Parliament passed by King Henry IV which would commit an d earrowsmith to jail should his products be found to be soft (Starley 2000: 179). h s i l There has been a limited degree of testing of this type on extant arrow heads. bReferring to small broadheads, Jones says ‘The blade is always hard, typically 350 u PVickers Hardness Number’ (Jones 1992: 112). Practical tests by Starley on period crossbow and arrow heads found levels of hardness ranging from 100–250 Hv (Starley 2000: 178–186). Testing by Bourke, Whetham & Stretton have found hardnesses of 105–158 Hv in a larger, lozenge-shaped medieval arrowhead (Bourke, Whetham, Stretton 2005). However, there is very little data on long ‘Type 7’ bodkins. The process used to manufacture these arrowheads suggests that they could be quite hard, especially at the tip due to the forging process. Test equipment
Bow: The heavy draw-weight bow used was hickory backed ‘compass’ yew as opposed to being self yew. This is a concession to reliability as a result of the
Defence Academy warbow trials
s e i r u o m r A e h t f o s e e t s u r T e h T ) c (
Figure 2
59
Mark Stretton with his 140-lb bow.
g scarcity of suitable quality yew. It has a draw weight of 140 lb at 32 in and has a n strung length of 80 in. This is a well shot bow and has come down from a ‘new’ i h s i l weight of 160 lb. Jones used a bow with a draw weight of 70 lb at 28 in. This was b u a modern style of longbow (figure 2) influenced by Victorian designs and P y shooting technique (stiff handle for a forgiving loose upright stance, straight draw e to the face as opposed to a ‘full compass bow, canted stance draw past the face). n a Current historical opinion, backed up by the Mary Rose findings, support a M y weight of between 90 and 150lb and a drawlength of around 30 in. b d Modern bow string materials were used in preference to a more traditional silk/ e h hemp string. It is felt that this is a justified concession to practicality. Considering s i l the draw weights and arrow weights being used, any benefit afforded by a b u lightweight string material is likely to be so small that it can be ignored. If these P
tests were using a modern target recurve bow with light weight and quick-acting limbs then string weight would be of importance. Arrow shafts (figure 3 a–c) are 31.5-in-long aspen, 7 in Turkey pinion flights, string bound, 1-in horn nock insert in accord with the British Long-Bow Society standard arrow based on findings from the Mary Rose (British Long-Bow Society 2001). Ash is considered to be an ‘ideal’ arrow-shaft material in terms of strength to weight and durability. The use of aspen was not expected to perform significantly better than ash and is still a historically correct material. The arrows used by Jones were roughly 2/3 of the weight of these arrows and, as such, the arrows used here will carry significantly more kinetic energy than those used by Jones. A number of identical arrows were made for the purposes of this testing. All were shot before the test to confirm their consistency.
60
Bourke and Whetham
s e i r u o m r A e h t f o s e e t s u r T e h T ) c ( g n i h s i l b u P y e n a M y b d e h s i l b u P
Figure 3a–c
Arrow shafts
All arrowheads were made by experts with a large degree of professional experience. The heads were made from Victorian iron. The lozenge arrowhead was intentionally hardened using a traditional technique of heating in a pot of bonemeal. Lozenge heads like these were in use from the end of the Hundred Years War and were employed throughout the Wars of the Roses. They are similar in appearance to heads commonly found on crossbow bolts (figures 4–6).
Defence Academy warbow trials
61
Figure 4 Head 1: Long bodkin (similar to ‘Type 7’), 71 g, 31.5 in from nock to start of arrow head. Ash shaft; bobtailed (1/2 in head, 3/8 in at nock). Vickers hardness: 190–200 Hv (Tip is harder, approx 300 Hv), Head Araldited on. This was the arrow head type originally employed by Jones.
s e i r u o m r A e h t Figure 5 Head 2: Short bodkin (similar to Type 10), 70 g, 31.5 in from nock to start of arrow f head. Vickers hardness: 230–250 Hv, aspen shaft; bobtailed (1/2 in head, 3/8 in at nock). o Arrowhead attached with hotmelt glue (see text). s e e t s u r T e h T ) c (
g n Figure 6 Head 3: lozenge, 87 g, 31.5 in from nock to start of arrow head. Vickers hardness: i h s 480–500 Hv (hardened in bonemeal), aspen shaft; bobtailed (1/2 in head, 3/8 in at nock). The i l b lozenge head is a long, heavy diamond bodkin. Arrowhead attached with hotmelt glue. u P y e Hot-melt glue was used to allow testing to continue should the arrows or arrow n a heads be damaged. The main heads used for the testing (lozenge and short M y bodkin) were kept as a constant and a number of identical arrows were available b d for maintenance. The use of a hot-melt glue as opposed to a more conventional e h hard-setting glue raised issues which will be discussed later. s i l b u PArmour
The initial thicknesses of armour were chosen to reflect those used by Jones, a flat plate is employed as it is somewhat more scientific than shooting at a breastplate (where the angle of impact obliquity can vary wildly making consistency difficult). The iron available today is generally of Victorian provenance. In general, this material is of a poorer quality than that which would have been used in the 14th century as it was mass produced with low emphasis on quality. Higher-end material (such as charcoal-rolled iron) is somewhat more refined as it is closer in quality to medieval iron. It was important that some of this expensive material was tested to allow the results to be as relevant as possible to medieval materials. Jones annealed his armour plates. The authors believe that it would be detrimental to the performance of a metallic armour system to be in a softened
Bourke and Whetham
62
state rather than a hard one. The likelihood of any medieval armour intentionally softening his product seems unlikely and as a result of this, all plates tested in this work will be in as supplied condition (figures 7–9).
s e i r u o m r A e h t f o s e e t s u r T e h T ) c ( g n i h s i l b u P y e n a M y b d e h s i l b u P
Figure 9a
Figure 7
Metal plate 1.
Figure 8
Metal plate 2.
Plate micro ID, b Vickers machine.
Defence Academy warbow trials
63
Metallurgy of target plates
The thinnest plate tested also appears to be the hardest. This is thought to be due to working of the metal by rolling to final thickness. Conventional ballistic armour theory suggests that the harder the armour is relative to the projectile the better it is. Therefore, metal plate 1 (figure 7) is the ‘best quality’ material of that tested. Microstructure s e The aim of this section is to observe the size and structure of the iron comprising i r u o the target plates. By performing microscopic examination any quality issues such m r as slag inclusions etc will be visible. A e h t f Method o s e Small samples from each material tested were taken and encapsulated within a e t s bakelite cylinder (figure 9). The samples were encapsulated in such a way that u r Tobservations could be performed on the flat, struck face. The thickest material e h tested was also included in bakelite in such a way that the through thickness Tstructure could be examined. Once encapsulated, the bakelite cylinder is ground ) c (
till the samples are flat and flush with the cylinder’s surface. After this, the surface
g n is ground with a diamond suspension abrasive fluid down to a surface coarseness i h of 3 µm. Following this the surface is polished and etched in ‘Nital’, a nitric acid s i l b based mixture. The result of this process is that the grain structure and internal u Pfeatures of the material is shown in sufficient contrast to be observed. y e Thin puddle iron (figure 10) has a reasonably large grain size; there is n a some slag distributed throughout the iron with some localised concentrations. MCharcoal-rolled iron (figure 11) has a small regular grain size, some slag y b d e h s i l b u P
Figure 10 Thin puddle wrought iron* 1.15 mm, good quality Victorian provenance, Vickers hardness: 206 Hv (max 221, min 191), microstructure: large, irregular grains, slag inclusions. *Iron ore smelted in coke furnace to cast iron, then furnaced and reheated to remove impurities through stirring (about 98% pure iron with slag). (left x20, right x50).
64
Bourke and Whetham
s e i r u o Figure 11 Charcoal-rolled wrought iron 1.95–2 mm* , probably similar to medieval quality, about m r A 99% pure iron with slag. Vickers hardness: 180 Hv (max 187, min 170). Microstructure: fine, regular grain structure with few slag inclusions. Good quality, strong material. *This would e h t probably represent some of the thickest parts of the breastplate. f o s e edeposits. Carbon content estimated below 0.1%. In thick puddle iron t s u r (figure 12), there does not appear to be much difference in grain size between Tthis plate and the thinnest plate. However, there does appear to more slag. e h T ) c Composition (
g n i hThese spectra, taken from samples of the target plates, give a general assessment s i l of the material used for testing. More detailed investigation is possible, however, b uat this stage, the general quality of the target iron is of interest. P The overview of the thin puddle iron (figure 13), shows that it is fairly y e nclean and free from significant amounts of slag, the presence of trace amounts of a Mphosphorus can be seen. Investigation of slag deposits has found typical amounts y b d e h s i l b u P
Figure 12 Thick puddle wrought iron 3.25 mm. As 1.15 mm plate but of lesser quality (required numerous attempts to get good hardness readings due to inconsistencies), Vickers hardness: 172 Hv (max 182, min 163). Microstructure: large amounts of slag inclusions, the through thickness structure of the material is lamina in its appearance.
Defence Academy warbow trials
65
s e Figure 13 Thin puddle overview. i r u o m r of silicon, phosphorus, sulphur and small amounts of manganese. Some small Airon oxide deposits were also found at grain boundaries indicating that this may e h t be recycled material or have been made in a dirty environment. These deposits f o also had large amounts of silicon and calcium which are likely to have been s e introduced through the smelting process. e t The overview of the charcoal-rolled iron (figure 14) shows it to be cleaner s u r than the puddle iron previously discussed. There is a trace of silicon visible but T e otherwise there are few impurities. The carbon content is notable. Carbon is the h Thardest element to detect using this technique and as can be seen is significantly ) more prominent than seen in the puddle iron. c (
g From the overview spectrum of thick puddle iron (figure 15), it can be seen n i h that the presence of silicon and phosphorus impurities indicates the presence of s i l slag. The slag deposits were in general very dirty containing very significant b u amounts of silicon, phosphorous, calcium, manganese, oxygen (in the form of P y oxides) and even chlorine. e n a MTesting and results y b d e Jones made no allowance for the armour to be supported by a body as if it h s was being worn: ‘No allowance was made for ballistic resistance of flesh because i l b the medical advice was that it is extremely small’ (Jones 1992: 115). This u P
Figure 14
Charcoal rolled overview.
66
Bourke and Whetham
s e Figure 15 Thick puddle overview. i r u o m r Astatement was made on the basis of a single private communication with a profes esor of forensic medicine but it appears to have been applied without taking h t f into account the context of the experiment. The ballistic resistance of flesh itself o s relative to that of the plate is indeed small. However, what is not accounted e efor is the support that flesh would give a plate (see above). The clay backing, t s ‘Plastalina’, being used in these tests is the closest that can be obtained to a u r Thuman torso (short of using prohibitively expensive instrumented crash test e hdummies) and is an oil-based flesh-simulating clay used in modern day police T ) body armour testing against ballistic and knife threats. The compliance of this c (
clay models that of the rib cage as a whole and has a similar resistance to flesh.
g nObviously, a strike on bone will not be simulated by this arrangement. This clay i h s is a standard simulant developed for body armour testing and to be valid for the i l bpolice testing standard (PSDB), this backing has to be used at an elevated u Ptemperature, in this case 35°C, to provide the correct degree of resistance. There y efore, the blocks need to be changed after spending approx an hour outside of the n aheating oven as their temperature (and their compliance) will change over time. M Target plates were mounted on 90 mm depth of Plastalina at 35°C in a y b wooden, backless frame (the backed steel box was quickly discarded as a test item d e hafter it was found that arrow penetrations were deeper than the box thickness, s i l hence the arrow heads were striking the back of the box and giving false b ureadings). Once the arrow velocity had been clearly established for each arrow P
type, the remaining figures were taken as representative rather than reconfirmed at each test. Impact energies were calculated using the standard equation: KE=½mv2 where KE is kinetic energy (J), m is mass (kg), v is velocity (m/s). Arrow 1 (long bodkin) was the first arrow retired from the test (Table 3) due to repeated damage to the point caused by failure to penetrate. As the least effective arrowhead type, this was not a significant issue. The shattering of arrow 2 (short bodkin) against the 2-mm plate at 60° was near the end of the day’s shoot and it had already performed very effectively (although slightly outperformed by the lozenge). We did not determine whether or not it would have been able to defeat the 1.15 mm plate at 60° although the authors are confident that it would have achieved this along with the lozenge, based on previous performance and similar characteristics.
67
Defence Academy warbow trials
s e i r u o m r A e h t f o s e e g e n a t s r u r m T 0 e 1 h t T a ) w o c b (
r
, n o i t a r t e n e m P m , y g r e n e c i t e n i K J
, y t i u q i l b o f s o e e e l g r n g e d A
g a n W i h : s , i l 1 s t s b s e u e n P T k c y i e 3 h n l e t a b t e M a g y T r a m b T m d e h s i , l y b t i u c P o
l s e / V m , t h g i e W g
d a e h w o r r A
n i d e n i a m e r 1 1 d 1 0 n 9 + + 1 1 1 6 7 1 5 4 5 a 0 0 0 0 0 0 2 3 9 6 2 5 5 2 4 7 3 2 4 6 3 5 5 8 2 0 8 1 d 0 0 0 0 0 0 8 9 1 8 9 1 8 8 / 8 1 e 1 1 1 1 1 1 - 5 t a r t e n e p e k i r t s d n o c 5 6 2 5 6 2 5 6 6 2 5 6 2 6 2 6 2 6 2 6 2 6 2 6 2 2 2 e 7 8 9 7 8 9 7 8 8 9 7 8 9 8 9 8 9 8 9 8 9 8 9 8 9 9 9 S . n o i t a r t e n e p s u o i v e r . p t l y u b s e r d e e s h u t 0 0 0 0 0 0 0 0 0 0 0 0 0 0 a 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 2 2 2 2 4 4 4 4 6 6 c o t e t t e a l c c t p a n c p e r a n e i p m f i f i n m f d i o o i y t e o a b l n ) ) ) d g m n r h y y y e n ) e l l l t r a i o k p f e n n n c t e w i o o o o t o t d h a d p t e y y y e i h o t a a a s t u c l l l 5 5 5 m r 5 5 5 5 5 5 5 d c ( c ( c e w w o ( 1 1 1 1 1 s m f e o . 1 . 1 . 1 . 1 . 1 . . . . . l r o t o r l 0 0 0 1 1 1 2 2 2 2 3 3 3 1 1 2 2 1 1 2 2 1 1 2 2 1 2 0 t r f r o e c 9 e a h a l ( k y e c t d d h r i y o e e n a s l v s a p t c w t a n u s a n h o e o o a e o t i s x r h a t e a t r d w l t 8 8 8 8 8 8 8 8 8 8 8 e p a f A e i r w c k t o 6 6 6 6 6 6 6 6 6 6 6 . k i . . . . . . . . . . . e t t n g e r 6 9 6 6 9 6 6 9 9 6 6 9 6 9 6 9 6 9 6 9 6 9 6 9 6 6 6 e i t n n u u o d c u 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 i o s i r e c h p t s w b a t , i t u e r t n b t o i n e u e , n w h t t , k d t o h t p u s c n o . d , i k o t a m r e d e d e c e e d a s h t l h t e p 1 0 7 1 0 7 1 0 0 7 1 0 7 0 7 0 7 0 7 0 7 0 7 0 7 7 7 t r n t t a t i a c s o 7 7 8 7 7 8 7 7 7 8 7 7 8 7 8 7 8 7 8 7 8 7 8 7 8 8 8 o u u r n , c t t t o d w e s b e e t l r u h t o n i a i d n s e f b l g i a e t u o d p n n p f a a o p n n n n n n n n n n n n n n n t s o i i i i i i i i i i i i r , e d e s u e i i i p h a t t h h k k k k k k k k k k k k k k k t t e u u t o r b e u d h d d d d d d d d d d d d d d d r o o d c o o e o o e o o o e o o e o e o e o e o e o e o e e e d o e d s a t d d d e b b g b b g b b b g b b g b g b g b g b g b g b g g g e t d l t d e t a e n n n n n n e n n n n n n t t t t t t t t t t t a a a e e e w t g r e g r e g r r e g r e r e r e r e r e r e r e e e r r e r i c c c t t m s n o z n o z n o o z n o z o z o z o z o z o z o z z z t p d n n n t e a e i t l o h o o h o o h h o o h o h o h o h o h o h o h o o o e e n r n h u u n u a L L L L L L L L e o o r o o g L S L S L S S L L S L S S S L S S S L r e c e T H A B B a S P t P 1 2 3 1 2 3 1 2 2 3 1 2 3 2 3 2 3 2 3 2 3 2 3 2 3 3 3 P B 1 2 3 4 5 6
0 7 8 9 1
Bourke and Whetham
68
Table 4
Arrow head
2 2 2 3 3 3 s 3 e i r 3 u o3 m r 3 A3 e h3
Short bodkin Short bodkin Short bodkin Lozenge Lozenge Lozenge Lozenge Lozenge Lozenge Lozenge Lozenge Lozenge
Test 2: Laboratory1
Weight, g
Velocity, m/s
Target thickness, mm
Kinetic energy, J
Penetration, mm
70 70 70 87 87 87 87 87 87 87 87 87
50 49.2 47.8 42.87 42.1 44.4 43.84 43.84 43.24 43.24 43.24 47.76
1.15 1.15 2 1.15 1.15 1.15 1.15 2 2 2 2 2
87.5 84.7 80 79.9 77.1 85.8 83.6 83.6 81.3 81.3 81.3 99.2
802 80 –3 70 70 904 71 125 36 107 107 137
t f o1For these and all subsequent shots, absolute max velocity measurement error=2.6% s 220 mm less than achieved by the bow although velocity and therefore kinetic energy was higher e e3No penetration, bounced out and dulled tip. Bow had same result except managed to penetrate 9 mm t s 4Suspect result as the arrow struck very close to previous hole in the plastalina clay u r 5Penetrated 12 mm but bounced out. Bow managed 16 mm and remained in target T 6Dented plate and bounced off. Not seen as representative as arrow sabots believed to be wearing and e hleaking air so replaced T7Dented plate and bounced out. Bow managed 16 mm and remained in target ) c (
g nRepeat of penetration tests in laboratory i h s i l bFor test 2 (Table 4), sponge sabots were placed around the nock end of the arrow u Pto provide a seal for the compressed air canon. Efforts were made to ensure that y ethis provided a ‘push’ force from the correct part of the arrow. The nock was n aflush with the back of buffer with a pin used to equate to string. The front bung M y was 70 mm from the front of the socket. b d e hDiscussion s i l b u Metallurgy P
The hardness of the plates tested in this experiment falls in the lower half of the hardness range of period armours which have previously been tested. This is especially the case when compared with later medieval and early modern armours where even the hardest of the plates used here would be softer than the softest of Henry VIII’s armours. This is important as the hardness of an armour is one of its prime methods of defeating a projectile. Jones annealed his plates to the fully softened condition and it is therefore anticipated that his plates were softer than the plates used in this work. The 2 mm thick plate tested was ‘charcoal-rolled’ rather than puddle iron and therefore had a far finer and more regular grain structure. This is better than some of the grain structures seen in some 15th century and earlier plates but at the same time worse than that seen in some later pieces. This plate is also
Defence Academy warbow trials
69
not especially hard. Jones comments on seeing spall from his armour and this is interesting as his plates were annealed and therefore relatively soft, a condition which does not lend itself to spallation. There was no evidence of spall from the materials tested in this work. However, due to the backing it was not possible to see any spall from the rear face. The micrographs of 15th and 16th century breastplates in Jones (Jones 1992: 114) and Starley (Starley 2000: 181) show a microstructure which is somewhat finer than the iron used in this trial. This suggests that the penetrations achieved against the armour thicknesses tested would be less impressive against better quality medieval armour. While starting from the position that much of the s armour that has survived to today is the ‘cream of the crop’ and that much e i r munition or low-grade armour has been lost to time or recycled, the authors also u o accept the observation made by eminent metallurgist, Alan Williams, who noted m r Athat even the best plate tested here is only of munitions-grade 15th century e armour and that Milanese suits of this period would have been of substantially h t better quality, accounting for their popularity (Williams 2006). f o The arrowheads used are of varying hardness but, in general, it appears that the s e hardness of the modern replica arrowheads is slightly greater than the period e t s pieces. The hardness of heads 1 and 2 is equivalent to the very hardest of the u r Theads tested by Starley (Starley 2000: 182–184). Head 3 is especially hard but e this head was intentionally surface hardened. The difference in these findings h Tmay be due to the fact that any surface hardening of the period arrowheads tested ) c has been lost due to corrosion over time. Jones comments that the hardness of ( g the blade portion of the small broadheads was 350 Hv and one might surmise n i h that hardened heads for defeating medieval armour may have been at least as s i l hard (Jones 1992: 112). As a result of his conclusions, Jones heat-treated all his b u arrowheads to 350 Hv, significantly harder than the heads used in this trial. From P y the investigations of period pieces, this is considered to be sound practice as e n long needle bodkins require a lot of working to achieve their final shape. This is a Musually accompanied with an increase in hardness in the highly worked areas due y b to the resulting fine grain structure. It is assumed that designs which required less d work will consequently be softer (and also therefore less brittle). e h s i l Penetration tests b u PSeveral of the arrows
on impact bounced out of the target plate or achieved poor penetrations. It was observed that, after these impacts, the socket of the arrowhead had opened up and the arrow forced into the head. It was concluded that the hot melt glue (used to allow quick changing of arrowheads) was too soft for purpose. There was general agreement amongst the test team that no examples of period arrowheads had been found with this sort of socket damage. The arrowheads were re-attached using an epoxy resin glue which gave a stronger joint and no further sockets were forced open. Penetration was then improved with those arrow heads. Stretton has completed some interesting tests on this area and concludes that the kinetic energy stored in an arrow is normally transmitted directly to the head. Where no glue is present, the socket is more likely to slip up the taper of the shaft, forcing the socket open and taking energy away from the arrow’s attempts to pierce and drive through the plate (Stretton 2006).
Bourke and Whetham
70
Against the thinnest plate tested (1.15 mm, i.e. 0.15 mm thicker than the plate Jones used) penetrations were more than double that which Jones recorded (for all heads tested). Against the medium plate (2 mm, the same as Jones but of significantly better quality charcoal-rolled iron), Head 1 (long bodkin) failed to penetrate unlike Jones’s work where 11 mm of penetration was achieved. This is expected to be due to two factors: firstly, the target plate was significantly better than that used by Jones and, secondly; the use of a heavier bow and faster arrow speeds overpowered and buckled the arrow head rather than penetrating the armour. Jones’ Long bodkin had approx 60% of the kinetic energy (KE) of the long bodkin used in these tests (the rest of the arrows used here have approxi s emately double the KE of Jones’ arrow). The failure of the arrowhead occurs i r uwhen a slender column is loaded in compression (such as a needle bodkin with a o normal impact against plate) as it has a tendency to buckle. A column will m r Ahave a ‘critical buckling load’ below which it will not buckle and fail (and thus econtinue to apply force to the armour leading to penetration). Above this load, h t f the column will buckle and as soon as this occurs, the strength of the column is o s massively reduced and the column fails rather than penetrating the armour. As e esoon as an arrow strike is not perfectly normal to the plate, the head will buckle t s far more quickly. At a range of 10 m, the yaw of the arrow due to the ‘archers u r Tparadox’ effect is still strong enough to cause a non-normal impact (figure 16). e h However, the correct heads for piercing of plate armour performed well, T ) with the lozenge (Head 3) performing the best although the really significant c ( penetrations still only occurred against the 1.15 mm plate. Against the thickest g nplate (3 mm) neither Jones’ tests nor those detailed in this paper succeeded in i h s i l penetrating the armour. b As discussed above, the thickest part of a breastplate is likely to be around u P2 mm though this is variable (see Table 2 for artefact A.22 in the Wallace y ecollection, no part is thicker than 1.5 mm). If the breastplate is made of particu n alarly good iron, the penetrations achieved in their own right are unlikely to be M y fatal. If the breastplate has thinner regions, or indeed is of a thinner metal all over b (as plenty of examples are) the penetrations recorded in this work would certainly d e hprove disabling or fatal. Despite the possibility of a non lethal arrow strike on s i l a thick breastplate, a heavily armoured solider brought to the ground by a non b ulethal arrow impact in a muddy, chaotic battlefield would find his chances of P survival severely impaired. Additionally, the energy carried by the arrows tested is so significant that even a non-penetrating impact in the right place might be sufficient to cause death by blunt trauma due to internal injuries (see Table 1).
Figure 16
Type 7 headed arrow yawing as it approaches target plate.
Defence Academy warbow trials
71
The data recorded during these tests corroborates the figures quoted by Williams very well indeed. The Health and Safety Executive lists an impact of 80J as a level of energy sufficient to cause death by blunt trauma (i.e. a non penetrating impact) (Health and Safety Executive 2002). Whilst a breastplate is likely to dissipate this somewhat there is still a good chance of a serious injury from a non-penetrating impact. Against the 2 mm charcoal rolled iron plates there is very little deformation around the impact site indicating that there is relatively little energy absorbed by the impact, rather it is transferred to the breastplate and therefore on to the wearer. This energy is obviously spread over a fair area and the type of padding or undergarment worn may also have a significant effect, s e however it evidently could potentially still be dangerous. i r u o Repeat of penetration tests in lab m r AIt is interesting to note that the attempt at re-creating arrow penetration in a e h t lab environment has so far failed to accurately simulate real world testing. f o For example, experiments conducted for a the 2003 series Battlefield Detectives s e involved ‘dropping’ the arrow head onto sheets of metal in an attempt to simulate e t s arrow strikes against armour (Granada 2003). Unsurprisingly, these tests ‘proved’ u r that the longbow was ineffective, ignoring the fact that the wrong arrowhead was T e employed (a long bodkin rather than the short armour-piercing bodkin found on h Tthe battlefield), the armour was backed with a solid piece of wood rather than ) c something that could simulate a person, and that simply dropping an arrowhead (
g onto a metal plate in no way replicates the action of the bow, even if the same n i h energy levels can be achieved in this way. The Defence Academy test team s i l wanted to see if it was possible to provide a more realistic laboratory test that b u would at least take into account the above factors. P In this spirit, it was decided for the tests to employ a compressed air cannon as y e n it could be calibrated to reproduce the same velocity consistently while at least a Mallowing the arrow to ‘fly’. However, even employing this technology, the arrows y b consistently failed to achieve the same degree of penetration that the bow d propelled arrows managed. There was a small degree of velocity error – the tests e h with Head 3 (lozenge) were on average 3 m/s (10 ft/s) slower than the velocities s i l recorded out of the bow, however this is small error (less than 6%) and is not b u Pexpected to be wholly responsible for the differences seen. This is backed
up with the tests using Head 2 (short bodkin) where speed error was close to 0.5 m/s (1.6 ft/s). Here the same reduced penetration for the lab tests was recorded. Whilst the air cannon trials replicate the arrow speed to sufficient accuracy it does not replicate a bow propelled arrow in terms of acceleration characteristics, the flex of the arrow is not the same and nor is the axial rotation of the arrow due to the spin stabilisation of the fletchings. An interesting phenomenon recorded by the high speed camera was that bow propelled arrows rarely struck the target straight and square — there was often a visible degree of yaw to the arrow. This yawing is due to the effect known as the ‘Archers Paradox’, caused by the simple fact that the arrow has to travel around the bow stave. Whereas the string returns to the centre of the bow, obviously the arrow has to go past the bow to continue
72
Bourke and Whetham
s e i r u Figure 17 Type 7 headed arrow fired from cannon not yawing as it approaches target plate (the o front sponge sabot can be seen preceding arrow). m r A e h t on to the target. At first, inertia causes the arrow to buckle while the bow pushes f othe arrow head to one side. The arrow shaft begins to vibrate and it is here s ethat it is so important that the arrow has the correct spine (flexibility) so that it e t s can recover from this and straighten out in flight as soon as possible. With heavy u r draw weight bows (say over 90 lb) matching spine is of less importance than the T eneed for an arrow to ‘stand-in’ the bow (ie be strong enough to withstand the hconsiderable forces applied to the arrow) As a result the bending of the shaft and T ) paradox is reduced because of the necessary stiffness of the arrow, however a c ( gyawing will still take place because of the effect of shooting around the handle n i h(Greenland 2001: 2–3). At which point this deviation in ‘clean’ flight dies out is s i l yet to be determined by further testing. b u In contrast to this effect, all arrows propelled by the air cannon travelled P y straight and struck the target square (figure 17). It appears valid to conclude that, ewhilst counterintuitive, the angle of strike not being exactly 90° might actually n a contribute to the effectiveness of the arrowhead penetration in some way when M y combined with the acceleration profile, spin and oscillation. Clearly, more tests b dare required on this phenomenon. e h s i l bConclusions u P
The longbow tests carried out by Jones in 1992 provided an important reference point for debates about the effectiveness of the medieval weapon. The intention of the 2005 Defence Academy Warbow Tests was to bring Jones’ tests up to date with contemporary opinion regarding the type and power of the medieval bow, weight of arrow, type of arrow head and the way the target itself was supported. It was then attempted to recreate these results under laboratory conditions. Metallurgical examination of the Victorian iron plate available for modern day testing indicates that it is of poorer quality than medieval plate. Surviving armour in general appears to be somewhat harder than the plate available to test. Charcoal-rolled iron plate is a better representation of better quality medieval armour although it would still not compare with the best Milanese armour. Modern replica arrowheads appear to be a fair representation of good quality original
Defence Academy warbow trials
73
arrowheads although it is unknown how hard average period arrow heads were or whether they would have been routinely surface/case hardened with the additional work this would have entailed. The replica arrowhead which was intentionally hardened using traditional methods is significantly harder than those period pieces tested. The techniques required to harden metals to this extent were certainly known about at the time so this could be due to surface hardening being lost (perhaps up to 1 mm) over time due to heavy corrosion. Against thinner plate (~1 mm), likely to be found in many areas of a suit of armour, penetrations of 80 mm or so into flesh can be expected with any of the arrowhead types tested. Against thicker plate (~2 mm), likely to be found on the s e front of the breastplate, penetrations achieved are unlikely to be fatal in their own i r u right; however, the energy of the impact may still be lethal (further tests are o required). Against thick plate (~3 mm), likely to be found only on the thickest m r Aparts of the breastplate and helmet, penetrations are unlikely. The effect of the e armour quality on the penetration performance is something that deserves more h t f tests. However, for the thinnest of the plates tested here, this factor, in the o s authors’ opinion, is of less significance than the thicker plates simply due to the e e huge degree of overmatch. After the initial penetration, the shape of the arrow t s u r head means that there is little arrow/armour contact until penetration reaches up Tthe socket of the arrowhead. By the time the penetration has reached most of e h the way up the socket, the hole in the armour will be almost fully developed and, T ) as such, the only influence of iron on slowing the arrow will be due to friction c ( g between the shaft and the plate. n The long bodkin arrow (Head 1 / Type 7) is effective against thin armour; i h s i l however, as the thickness increases, the effectiveness of this arrowhead reduces b u rapidly until a point at which it fails by buckling rather than penetrating. Jones Pactually achieved better results using this type of arrow head with a lighter arrow y e shot from a lighter bow and it is believed that a heavier bow just overpowers this n a type of head. The short bodkin (Head 2 / Type 10) performed significantly better M y than the long bodkin against metal plates, either in this test or in the original b d Jones 1992 tests, and demonstrated the ability to punch through to a lethal depth e h against thinner plate at an oblique angle of at least up to 40°. The lozenge-shaped s i l head penetrated the thinner plate even at an extreme oblique angle of 60° (if the b u test arrows had survived, it may have been possible that the short bodkin would P also have been able to achieve this degree of penetration). Clearly, both the short bodkin and the lozenge arrow heads performed significantly better than the results achieved back in 1992. Arrow heads that were securely glued on to the arrow shaft outperformed those that were merely hotmelted on and it was also clearly established that war arrows loosed from a heavy bow possess a significant amount of energy and are theoretically capable of killing by blunt trauma alone should enough energy be applied to a critical area. Some questions arose as to the distance of the test. At a range of 10 m, the arrow flight has not fully stabilised and therefore this may have a detrimental effect on impact performance. However, the decreased performance of the test arrows in the laboratory when shot at similar velocities may indicate that striking the target ‘square’ is less important than other factors such as arrow spin, stored
74
Bourke and Whetham
energy in the arrow shaft or even the effects of the Archer’s Paradox itself. This raises some intriguing questions (to be explored in future tests) that show that laboratory testing is not currently suitable for simulation of arrow impacts or for accurately gauging the effectiveness of the medieval warbow. Acknowledgements
The project co-ordinators were David Whetham (King’s College London), Paul Bourke (Cranfield University) and Hilary Greenland (bowyer and fletcher, SPTA). The test team comprised Mark Stretton (heavy bow expert, blacksmith s eand fletcher), Hector Cole (master arrowsmith and archaeological blacksmith) i r uand Hugh Soar (archery historian). Thanks are due to Roy King (consultant o m r armourer), Dr Ian Horsfall, James Shattock and the Bashforth Laboratory.
A e h t Notes f o s e1. The British Museum numbering system, introduced in 1940 and commonly used for e t the categorisation of arrowhead types, is now considered to be slightly misleading s u (Ward-Perkins 1940: 65–73). The Jessop numbering system, introduced in 1996, is now r T considered to be more useful (Jessop 1996: 192–205). However, to avoid any confusion e the arrowheads used in this work are described individually below with photographs with h T similarities to convention being noted. ) c 2. Williams quotes work by McEwen suggesting a 200-J limit for crossbow bolts based on (
experiments using a modern crossbow (Williams 2003: 919). No data is available for the g n i velocity of the Payne-Gallwey Genoese bow but using the weight data provided and h estimating a 50 m/s velocity (similar to our longbow) energy has been calculated by stan s i l dard formulae. Due to the 18-lb weight of this weapon, it is unlikely to have been common b u on the battlefield. P y e n aReferences M y bBourke, P, D Whetham and M Stretton 2005 Analysis of medieval arrowhead, un d published report, Defence Academy of the UK, Cranfield e hBritish Longbow Society 2001 Rules of shooting . 4th Edition., Annex B, 11. Rotherham, s i l Factandfiction b uCroft, J 2003 PSDB Body armour standards for UK Police, Pt. 2, ballistic performance . Publica P tion Number 7/03/B Home Office Police Scientific Development Branch
Curry, A 2001 The Hundred Years War, in Holmes, R (Ed) Oxford Companion to Military History . Oxford, Oxford University Press DeVries, K 1997 Catapults are not atomic bombs: towards a redefinition of ‘effectiveness’ in premodern military technology. War in History 4: 454–470 Gerald of Wales 1978 Journey through Wales and the description of Wales. Translated by L Thorpe. Harmondsworth, Penguin Greenland, H 2001 The traditional archer’s handbook: a practical guide . Bristol, Sylvan Archery Granada, 2003 Battlefield detectives: Agincourt 1415 . Grenada Television Product for Five and the Learning Channel. Hardy, R 1986 Longbow: a social and military history . London, Patrick Stephens Ltd Holmes, R 2002 (commentary in) Royal Armouries Arms in Action Series II: Bow. Yorkshire TV, Leeds. Jessop, O 1996 A new artefact typology for the study of medieval arrowheads. Medieval Archaeology XL: 192–205.
Defence Academy warbow trials
75
Jones, P 1992 The metallography and relative effectiveness of arrowheads and armor during the Middle Ages. Materials Characterization 29: 111–117. Jones, P 1995 The target . In Hardy, R Longbow: a social and military history . Appendix 3. London, Patrick Stephens Ltd King, R private communication with authors. Mary Rose Trust 2006 Electronic archive. www.maryrose.org/mary_rose_archive.html Monstrelet 1840 The chronicles of Enguerrand de Monstrelet Vol 1. Translated by T Johnes. London. Payne-Gallwey R 1995 The book of the crossbow . Dover Publications Rees, G 1993 The longbow’s deadly secrets. New Scientist , 5 June, issue 1876: 24 Soar, H 2004 The crooked stick: a history of the longbow . Pennsylvania, Westholme Publishing Starley, D 2000 Metallurgical analysis of medieval quarrel heads and arrowheads. Royal s e Armouries Yearbook 5. Leeds, Royal Armouries: 178–186. i r Stretton, M 2005 Practical tests — part 2. The Glade No. 108: 52–56 u o Stretton, M 2005b Fire arrows. The Glade No. 109: 16–20 m r Stretton, M 2006 Practical tests — Part 4. The Glade No.111 A Strickland, M and Hardy, R 2005 The great warbow: from Hastings to the Mary Rose . Stroud, e h Sutton t f UK Health and Safety Executive 2002 Controlling risks around explosive stores, review of the o s requirements for separation distances . London, HMSO. e Ward-Perkins, JB 1940 London Museum medieval catalogue . London, HMSO e t s Waurin, J 1868 Anchiennes croniques. Vol II ed. ELCP Hardy. London, Longmans u r Williams, A 2003 The knight and the blast furnace: a history of the metallurgy of armour in the T middle ages and the early modern period . Leiden, Brill e h Williams, A 2006 Private correspondence with the authors, dated 11/10/06. T ) c (
g n i h s i Comments l b Kelly DeVries u P Department of History, Loyola College, Baltimore MD y e Reply n a David Whetham and Paul Bourke M y b d The following comments by Professor DeVries on the above paper are addressed in turn e h by the authors and are published here (with references to the page numbers of the paper s i l and to the list of references at the end of the paper) in order to provide additional insight b u into the study. P
DeVries: As the authors say at the beginning of this article, several historians since have used the 1992 experiments conducted by Peter Jones to justify their own work. But they are also in need of review and updating, especially as the study of longbow archery use and effectiveness has produced some interesting — and sometimes controversial — results. Now I should say from the outset that I have a few problems with some of the methodology used by the team making this report, as will become somewhat clear below, but having said this I actually think that it is important to publish any new results, so long as they follow the rigour of these tests, thus allowing the debate over effectiveness to continue. The debate is here in this article, too, although the authors in stating their thesis indicate only a willingness to revisit Jones’s findings. I do have a slight
76
Bourke and Whetham
criticism of this, as by the time one reaches their conclusions it becomes clear that in fact they claim not only to have revisited Jones, but to have proved the lethality of longbow archery, even if no penetration of armour occurred. They also claim to have shown that the bows used by English archers at the end of the Middle Ages were very heavy, some 120–140 lb. This is not entirely surprising, as in using Mark Stretton and others who have biases towards these heavy pull bows, it might appear that such a thesis was preconceived before the tests began. Nor is it surprising in that the authors sometimes dismiss methodologies without detailed reasons, which, at least from the text at hand, seems to show somewhat manipulated results to support their preconceptions. Hence my criticisms. s eBourke & Whetham: The purpose of the 2005 tests was to revisit the excellent i r u ogroundwork done by Jones back in 1992 and bring it up to date. To do this, a m r team was assembled that would allow us to draw upon more than simply A eacademic experience. It consisted of experts in their fields as blacksmiths, h t bowyers, fletchers, physicists and archery historians. After reviewing Jones’s work, f owe agreed to use an appropriate heavier bow of the right design, the correct s earrow design and weight to match the bow (the bow and the arrows have to e t s be ‘matched’ or they do not work efficiently), the correct arrow head (rather u r Tthan an obsolete and mismatched head designed for a different purpose) against ea properly supported target plate. h T There were, of course, still some limitations to these tests. One of the main ) c ones being not having a real suit of medieval armour on a real person to shoot ( gagainst (of course, we would need a representative suit, in itself, a whole area of n i hdebate and several weeks to allow the consistency of the shooting in each section s i l bof the armour and allow for the repairs between each shot). The enormous prob u Plems associated with addressing all of the potential difficulties led us to accept y emuch of Jones’s original methodology — specifically, metal plates shot at 10 m, n athen angled to provide an increasingly oblique surface for accurate and consistent Mmeasurements. The metallurgical analysis of the target plates was then provided. y bOur second series of tests done in 2006 (to be published in the next edition d eof Arms and Armour) do employ a range of bows from 110–150 lb, shot at a h s i l number of differing ranges, to provide a broader spread of results, and we are also b uhoping to secure funding for another series of tests against some replica armour, P specially forged using appropriate methods and materials. However, back in 2005, we had to start somewhere! We are a little more concerned with the charge of manipulating results — much of DeVries’ criticism seems to be aimed at our assumption that heavier bows were the norm. The authors accept that there is a healthy debate about the effectiveness of the medieval warbow, but we are also surprised that people still contend that professional archers in the late middle ages might have been using such a light weapon. The Tudor period bows recovered from Henry VIII’s flagship the Mary Rose, have an average length of 78 q and estimated pull weights of 100– 180 lb (Pratt in Hardy 1986: 209–217). The argument for smaller bows is neither born out by the evidence (some more of which is addressed below), or common sense. Both of the authors (hardly the most athletic of specimens — Paul weighs
Defence Academy warbow trials
77
in at under 10 st!) happily pull a 70-lb medieval-style self-yew bow on the type of irregular basis that most amateurs today have to settle for. This is a heavy pull for us but we don’t do this professionally. With eight hours of daily practice, a good diet and the incentive of a continued salary, it is difficult to see why a professional medieval archer would stick to something this light when he had the capability of pulling a larger bow with its greater range and penetrating power. Our test archer, Mark Stretton, routinely pulls 150 lb or more. The technology does not get more complicated to make a bigger bow. Do people really think that someone capable of pulling a ‘real’ bow like this would be happy with such a 70 lb ‘toy’ when their livelihood (and life) depended upon it? In short, the bowyers feel these bows s e can be made, the archers demonstrate such bows can be shot and the historians i r u generally agree that the archers were well paid professionals rather than the o wizened and weak peasants portrayed by Hollywood. m r ADeVries: The statement at the bottom of p. 54 that ‘It is now widely believed that e the Mary Rose arrows are very similar to their medieval predecessors.’ needs a h t f reference. As the Westminster arrow is of similar size this may well be accurate, o s but without a date on the Westminster arrow and the late dates of the Mary Rose, e e I think the statement is too bold without some substantiation. t s u r Bourke & Whetham: Relevant references are Rees (1993) and Strickland & Hardy T(2005: 27). Moreover, why would the arrows be significantly different? The only e h extant archaeological evidence shows arrows of this length. If someone believes T ) that at the height of the weapon’s popularity it was less effective because the c ( g arrows used were smaller (thus indicating a smaller bow and the resulting reduc n tion in power, etc.), surely the onus is on them to prove this as this goes against i h s i l the extant evidence? b u DeVries: I appreciate the team using something to represent skin in these tests P(p. 56), but what about other clothes or padding? Armour was not placed on y e skin alone, and all layers need to be factored into the tests — also probably the n a potential movement of all of these layers in imitation of a man in battle. Again M y these omissions should not keep the findings from being published, but a reason b d why one layer was imitated but not the others might keep readers such as me e h from worrying about the incompleteness of the test targets. s i l Bourke & Whetham: The Plastalina was not for representing the penetrative b u resistance of flesh, but rather to assess the penetration of the plates themselves P when supported by a human torso. It was also hoped that we could determine the level of trauma behind the armour, but this proved difficult to assess so the team employed a Behind Armour Blunt Trauma (BABT) rig for the next series of tests. Additional layers of cloth protection beneath the plates may or may not make a difference. To determine this, further tests are required and we look forward to being able to do this. We also would love to know how to imitate the movement of a man in battle for future tests! DeVries: Is the softness of English armour (p. 56) relevant without a similar indication of continental European armour softness, especially as most English arrows were shot at those targets? Bourke & Whetham: Williams (2003) cites figures for European armour hardness. However, the point here was to demonstrate the problems of accounting for the
78
Bourke and Whetham
clear variance in thickness and hardness in different suits of armour, not to establish if one country’s armour was harder or thicker than another’s, an assertion that we doubt could be made authoritatively precisely due to this problem (although we would be delighted to be proved wrong). DeVries: Gerald of Wales is not a reliable source for judging the effectiveness of weapons (p. 57). He exaggerates almost all of what he writes – what do they mean embellish with time? One might as well use the Maciejowski Psalter as an accurate portrayal of combat wounds, or the Song of Roland for sword penetration (through a helmet, armour, and the horse, too). Gerald certainly can not be trusted. Far more sources talk about the effectiveness of mail, such as Joinville s ewho mentions arrows sticking out of every bit of the Crusaders’ mail but none i r upenetrating it or the layers below it. o Bourke & Whetham: Gerald of Wales probably does exaggerate, but one must m r Aalso not simply dismiss everything he says. While there are obviously many vari eables involved, practical tests demonstrate it is feasible to shoot a long bodkin h t f arrowhead 2q into seasoned oak. Not quite a palm’s breadth (taken to be about o s 3.5q ) but not too far out either (Stretton 2005: 52–56). Joinville is a fascinating e esource, but here he is talking about arrows shot from a short composite bow t s rather the type of bow we are talking about. Even allowing for the better armour u r Tavailable in the 15th century, at Agincourt Monstrelet recorded that ‘numbers of e hthe French were slain and severely wounded by the English bowmen’ (Monstrelet T ) 1840: 342). The eye-witness Jean de Waurin also recorded that many were dis c ( abled or slain by arrows before they could come to close quarters (Waurin 1862: g n213). i h s i l DeVries: Again there is the problem of substantiation when it comes to a state bment such as ‘it is believed that a point was reached at which plate armour u Pnot only offered better protection than mail but was also cheaper’.(p. 57) Who y ebelieves this? I have never read this, nor do Claude Blair or Alan Williams say it. n aA reference is necessary. M y Bourke & Whetham: We believe that given the economic facts this logically b dfollows combined with practical tests that demonstrate that riveted steel mail e hprovided little or no protection against any of the arrowheads, from crescent s i l shaped through to heavy bodkin, when they were shot from a heavy bow b u(Stretton 2005: 23). Obviously, plate can provide better protection than this. P DeVries: There are no historical references to fire arrows (p. 58) in the late Middle Ages shot by longbows. Hollywood likes them, but historians really need proof and something more than the obscure and inaccessible articles of Stretton in The Glade (Stretton 2005b). Bourke & Whetham: Rather than flights into Hollywood fantasy, this was simply a suggestion to try to explain the existence of an arrow that otherwise appears to be unusable. Used in this way, it does work — and very effective it is too! As to the work being inaccessible, does the reader mean hard to read or hard to find? Neither of these is a valid criticism given that The Glade is an international publication. Just because Stretton is not an academic does not mean that one can not understand what he is saying — a charge non-academics can rightly put to many of us!
Defence Academy warbow trials
79
DeVries: Again substantiation is needed when the authors write ‘Current historical opinion, backed up by the Mary Rose findings, support a weight of 90–150 lb and a drawlength of around 30q (p. 58) Strangely — or maybe not — when the authors claim to be citing ‘current historical opinion’ or ‘it is believed’ they are glossing over something that calls to question their conclusions or methodology. In fact, there are many scholars who consider this opinion to be erroneous: myself, John Waller, Valérie Serdon (her book, Armes du diable: Arcs et arbalètes au Moyen Age, Paris, 2005, should really have been consulted somewhere in this article). Bourke & Whetham: See above. s DeVries: Modern bow strings are not the same as traditional silk/hemp string e i r (p. 59), and they do not give the same results. For one thing, they do not break u o as regularly, and they do not need the same thickness as earlier strings — espe m r Acially important as the thickness is determined by the nock size, with a modern e string filling the nock certainly more powerful than its medieval equivalent. h t Again, this makes one question some of the results of these tests. Where is the f o s evidence to substantiate this claim? e Bourke & Whetham: Indeed, modern bowstrings are not the same as medieval e t s ones. From the size of the nock on the arrows that did survive on the Mary Rose, u r Twe can deduce that the strings could have been no wider than 1/8 of an inch e (3.2 mm). This is similar to the thickness of many modern strings once wrapped h Tin serving thread (a reinforcing thread added to the string at points of wear). ) c The nock size is a red herring as this would be matched to the string size gener ( g ally in use or it would be inefficient (fletchers were professionals who knew what n i h they were doing). If we knew how to make the strings and their protective resins s i l in the same way, we would. Again, it seems that to assume they were not as good b u as today is to take a worse case scenario to the extreme. The concession to a P y material that does not break as often appears to be both practical and fair. e n DeVries: Why is aspen not used instead of ash? Again the two woods are quite a Mdifferent and would produce different results, with aspen expected to perform y b less well than the harder ash. Also aspen was the standard wood for English d longbow arrows — see Henry V’s prohibition against using aspen for anything e h other than arrow shafts. More substantiation to support their claims here is s i l needed. b u PBourke & Whetham : Aspen and ash are both employed in the test. DeVries: Were any of the arrowheads steeled (p. 59)? Bourke & Whetham: We are not sure which definition of steeling the reader is using, but the test report clearly states where the arrowheads are case hardened and the materials used to do this. DeVries: Why are the authors confident that such an arrow (p. 66) would penetrate armour 1.15-mm thick when shot at such an oblique angle as 60°. Their findings do not suggest that they should support such a finding when there was no penetration on the 2-mm plate when shot at this angle. Bourke & Whetham: We are confident that the short bodkin would have penetrated the plate at 60° because, as the results show, it had consistently performed in a similar way to the lozenge. The lozenge achieved 80 mm of penetration at this angle.
80
Bourke and Whetham
DeVries: The assertion of better armour quality (p. 68) should probably be weighed against the Rhodes armours studied by Chip Karcheski and Thom Richardson (Walter J. Karcheski, Jr. and Thom Richardson, The Medieval Armour from Rhodes [Leeds: Royal Armouries and Higgins Armoury Museum, 2000]) before categorically claiming such. Bourke & Whetham: Perhaps it should. DeVries: I find it rather odd that on p. 70 and later the authors assert a lethality figure that has been increased due to their estimation of wounded or ‘knocked down’ soldiers dying from their injuries. This is unsupported by any evidence in the article (or actually elsewhere). In fact, the healed wounds found on the s eexcavated bodies at Towton and Visby would suggest that it was quite likely that i r ua soldier could survive his wounds, even to the head and even from arrows. o Meanwhile the idea that non-lethal arrow hits to armoured soldiers caused deaths m r Aat a high rate — as seems to be indicated by the authors – simply is not supported e by contemporary evidence. Finally, the Health and Safety Executive was unlikely h t f to have considered the very small impact of an arrow when determining their o s blunt force trauma index. e e I also question the use of a 1200-lb pull crossbow, which was unlikely to have t s u r been used in many military conflicts. This does, however, indicate once again a Tpreconception of heavier weapon use. e hBourke & Whetham: The tests demonstrated that, while there was no penetration T ) of the thickest plate, the thinner plate could be pierced to at least a depth of 80 c ( gmm even at the most extreme oblique angle of 60°! Due to the less critical areas nlikely to be covered in this thickness (although again, this not easy to substantiate i h s i l on a representative basis), this might not prove fatal, but the recipient of such a b uwound is unlikely to be taking an active role in the ensuing encounter, and being Pinjured or at the least knocked off one’s feet, would likely be vulnerable to further y eattack on a medieval battlefield, even if not by a longbow missile. The comment n aregarding the HSE is probably valid. However, under armour, the impact will M y likely be one of blunt trauma rather than point impact due to the spreading b dout of the force of the impact. Finally, Payne-Gallwey (Payne-Gallwey 1995) e hcertainly believed the 1200-lb crossbow he was shooting was of medieval s i l provenance. b uDeVries: On p.70, the authors begin to hedge their bets on the non-lethal impacts P being lethal because they finally and for the first time introduce padding and undergarments on the soldier. These have not been factored in elsewhere in their experiments. This also seems to contradict what they have already said about these wounds and what they go on to say over the next couple of pages. Bourke & Whetham: It is correct that we did not use padding. See above. DeVries: The section entitled ‘Repeat of Penetration Tests in Laboratory’(p. 68) is rather confusing. While criticizing the experiments of Battlefield Detectives — a criticism I wholly support — the authors then seem to rationalize some of the modern techniques they used in the lab, only to indicate some doubt of the methods at the end of the section with ‘clearly, more tests are required on this phenomenon’. One might suggest that this undermines the confidence in their research that they exhibited earlier in the article.
Defence Academy warbow trials
81
Bourke & Whetham: This is what happens when you test something! Had we wished to ‘prove’ a thesis, this would not have been difficult, but it would also not have been appropriate. We do not have doubt over our ‘real world’ results, we were simply attempting to recreate them in the laboratory. We were unable to do this despite matching the velocity of the arrows. We did not set out to prove it was impossible to replicate the results in the laboratory with this equipment. DeVries If the vibration of the arrow is important (pp. 71–72) to consider in the tests, then wouldn’t the wood used for arrowheads be important, too? The authors dismissed aspen above, despite the fact that it would have vibrated more than ash because of its density. s e Bourke & Whetham: We were merely commenting that the laboratory testing did i r u not replicate the (substantial and very visible on the slow-motion camera) flexing o motion of the arrow shaft achieved in the ‘real world’ and that this is something m r Aworth exploring further. e DeVries: What is the ‘contemporary opinion’ (p. 72) the authors are referring to h t f at the beginning of their conclusion? From recent work, I would say opinion is o s evenly mixed between those who believe a lighter bow was used and those who e e believe it was a heavy bow. Again, this indicates a preconceived thesis. t s u r Bourke & Whetham: The ‘contemporary opinion’ includes Strickland & Hardy T(2005), Soar (2004) and Holmes (2002), as well as the team of experienced e h practitioners involved in these tests. It is difficult to see what evidence the thesis T ) of a lighter bow actually enjoys in its favour. c ( g DeVries: Once again (p. 73) non-lethal wounds are said to be lethal, something n not proven by the tests above or any other study. As the authors seemed to be i h s i l considering that padding and undergarments might add protection on p. 56 b u (I would say considerable protection, adding to that of the movement of the body Pwhich is unconsidered by the authors), the introduction of lethality here seems to y e hearken again to a preconceived thesis. n a Bourke & Whetham: See above. M y b d Conclusions e h s i l b DeVries: As I u Pbe published.
indicated above, I do think that studies such as these should However, I remain unconvinced that they have achieved much to advance the argument. Usually, if one sets out in a laboratory to achieve a preconceived thesis, one will achieve it. Bourke & Whetham: We are confident that, alas, the eminent Kelly DeVries will remain unconvinced of the heavier bow thesis no matter how compelling the evidence presented. We will, however, continue to try. Part II of the Defence Academy series of tests conducted in the summer of 2006 will further add to the debate.
