Testing reconstructed medieval crossbows featuring composite bows by Andreas Bichler


1 Introduction

There are a few medieval crossbows equipped with composite bows in several European museums but they are in such bad condition that no shooting experiments can be made.

The following experiment tries to explore the effectiveness of such crossbows by testing reconstructed models on reconstructed armour typical of the 13th and early 14th century.

Basically it has to be said that this test cannot show exactly how effective original crossbows were, since reconstructions are only an approach to them. Too many details are likely to differ from the originals: For example, if we use the same kind of wood, we hardly know if the density of the wood is correct. What is more, we generally do not know the precise manufacturing process by which crossbows were built 700 years ago. In this regard it’s impossible to achieve results that are reliable by a 100 %.

However, tests such as the following help to learn more about the capability of medieval weapons. We can also receive a better understanding about the race between protective and attacking weapons, constantly influencing and developing themselves.


2 Basic conditions

2.1 Crossbows

The crossbow with a composite bow was brought to Europe from the East by Saracens, during the Crusades of the 12th century[1]. Between the 12th and the 15th century the crossbow was the favourite weapon of medieval armies at the Continent.

The reconstructed composite crossbows (fig. 1) based on pictures of 13th and 14th century, as well as on preserved pieces of the late 14th and 15th century and are made of authentic materials such as horn, sinew, deer antler, bone and wood.

Bows were dimensioned in order to require two different drawing devices: Crossbow 1 has a draw weigth of 131 kg, meaning this bow is drawn by a metal hook hanging from a waist belt. Crossbow 2 features a draw weight of 280 kg. This model requires a so-called windlass to draw. Windlasses were mainly used in the late 14th and 15th century.

When shot at an angel of 45 degree, these crossbows reach maximum shooting distances of 220 to 240 m (crossbow 1) and 260 to 300 m (crossbow 2) respectively.

It’s interesting that such distances can be found in medieval sources, e.g. the “Dunstable chronic” dating from the first half of the 15th century: Henry V. is described as approaching the city of Rouen (France) from within a distance of 40 rods (201 m) or within the quarrel shot of a crossbow[2]. A second example can also be found in this chronic: A crossbow archery on the ice-bound Bodensee from 1435 is described where 21 crossbow-shots reached a distance of  7 kilometres, meaning one shot made up for an average distance of 333 m[3]

Fig. 1 Reconstructed crossbow 1 and 2


Technical data 

Crossbow 1:              Bow length 75cm
 Draw weight
[4]: 131kg
 Power stroke: 206mm
 Overall weight: 2kg


Crossbow 2:              Bow length74cm
 Draw weight
[5]: 280kg
 Power stroke: 230mm
 Overall weight: 3kg


2.2 Quarrels (bolts) 

Four different types of reconstructed bolts (“quarrels”) were used. All wooden shafts are based on preserved pieces of Habsburg castle (Switzerland)[6]. These shafts were equipped with different heads according to findings dating from the 13th and 14th century[7]. On the one hand, slim lancelet heads were chosen, on the other heavier, blunt shapes.


2.2.1 Bolt 1 (Fig. 2)          Lancelet head with rhombic cross section at a shaft of larch wood with feathers of willow

                                               Head broadness: 9mm
            Head thickness: 5mm
nozzle diameter: 13mm
            Head length: 87mm
            Overall length: 381mm
            Mass: 41g

Fig. 2


2.2.2 Bolt 2 (Fig. 3):         Blunt pyramidal head with rectangular cross section at a shaft of beech wood and feathers of willow

                                               Head broadness: 10,7mm
            Head thickness: 10,6mm
nozzle diameter: 15mm
            Head length: 78mm
            Overall length: 381mm
            Mass: 63g


Fig. 3


2.2.3 Bolt 3 (Fig. 4):         Blunt pyramidal head with rhombic cross section at a shaft of larch wood and feathers of willow

                                               Head broadness: 13,2mm
Head thickness: 10mm
External nozzle diameter: 14,7mm
Head length: 95mm
Overall length: 392mm
Mass: 67g

Fig. 4


2.2.4 Bolt 4 (Fig. 5):         Blunt pyramidal head with rectangular cross section at a shaft of beech wood and feathers of willow

Head broadness: 15,2mm
Head thickness: 15,4mm
External nozzle diameter: 16mm
Head length: 87mm
Overall length: 389mm
Mass: 90g

Fig. 5


2.3 Targets

2.3.1 Target 1:       Represents the simulation of human body by using three ballistic-soap blocks with the masses of 25 x 25 x 40cm. Ballistic-soap (glycerine cream) is similar to gelatine. Both are used for making wound-ballistic experiments, although differences concerning penetration-depths between these two materials may occur, as Hubert Suedhus found[8].

2.3.2 Target 2:       Two blocks of ballistic-soap are now protected by reconstructed medieval textile armour (gambeson) (fig. 6). This gambeson consists of two layers of felt (each layer approximately 12 mm thick) (fig. 7), coated by one layer of linen and quilted in vertical lines, resulting in an overall thickness of 12 (in the quilt seam area) to 22 mm.

2.3.3 Target 3:      Additionally to target 2 a shirt of reconstructed medieval mail consisting of alternately riveted and solid iron rings is used. Ring diameter inside approx. 9,5 mm, wire diameter before flattening 1,3 mm (fig. 8 and 9).


Fig. 6: Textile armour

Fig. 7: Core of two layers of felt





Fig. 8: Shirt of mail over textile armour

Fig. 9: Alternately riveted and solid rings



2.4 Speed measurement

Speed measurements were taken by using the measuring instrument BMC 12, W. Mehl, Kurzzeitmesstechnik. Devices operating with light barriers were set up in order to achieve v1 (cf. 2.5). 

2.5 Shooting distance and configuration 

The test scenario was built up in a closed space and took place under a surrounding temperature of 25˚C. Shot distance was 10 m and speed (v1) was 1 m measured from the back of the crossbow’s bows to the target.


3 Results

Generally it must be stated that a major statistical rule, that is the more repetitions (in this case shots) are made the more exact the average value gets, must be ignored here, because the leader of the test didn’t want to put too much stress on the crossbow’s composite bows. Therefore, only a small number of shots was made, meaning the results are likely to differ more widely from an average value as usual.

Due to the low shot result the calculation oft the standard divergence was renounced, because the preserved values are valid as small tests and are valid, hence, with the calculation of variables as unreliable.

3.1 Overview target 1 

20 bolts were shot at target 1. The main hitting zone occurred on the middle block and spread to an area of 23 cm in width and 39 cm in height (fig. 10). Penetration depths (fig. 11 and 12) were generally in an area of 143 to 208 mm, resulting in an average depth measuring 170,5 mm.

Penetration depths however depend on bolts and weapons used. Thus, identical penetration depths can be observed in spite of different speeds and bolt weights. 

Fig. 10: Hits at the soap blocks


Fig. 11: Hit from bolt 1

Fig. 12: Hit from bolt 4


3.2 Overview target 2 

On testing target 2  penetration depths ranging from 71 to 103 mm were achieved (bolts 1-3) (fig. 13). Bolt 4 bounced two times off, however, left a deep dent of 18  an 20mm in the textile armour as well as an perfectly circular impact, 45 and 46mm wide and 16 and 18mm deep (fig. 14).


Fig. 13: Hit from bolt 1

Fig. 14: Deformation of the soap block after hit from bolt 4


3.3 Overview target 3

The combination of textile armour and mail (fig. 15) was fired 12 times. Only one type of bolt was able to completely pierce gambeson and mail: bolt 1, featuring the slim, lancelet head. This bolt penetrated between 68 and 83 mm and was only stopped by the mail rings in the area where the head ends and the shaft begins.

Three shots  were not scored, because they didn’t hit the area backed by the soap blocks. It was noticeable that bolts that impacted at an angle of approx. 90° only caused stretched rings (fig. 17 and 18), whereas impacts at steeper angles (e.g. 50°) resulted in broken rings (fig. 18, 19 and 22). While bolts 2 and 3 were absorbed by the mail rings and hardly damaged the textile armour, heavy bolt 4 completely bounced off and left only deformed rings (fig. 20 and 21).


Fig. 15: Hits on target 3

Fig. 16: Hit 8

Fig. 17: Detailed sight of hit 8




Fig. 18: Detailed sight of hit 2

Fig. 19: Detailed sight of hit 4




Fig. 20: Detailed sight of hit 11

Fig. 21: Detailed sight of hit 12





Fig. 22: Broken rings


3.4 Determination of the bolt speed 

Bolt speed was measured by two light barriers set up - 50 cm apart from each other - at a distance of 1 m from the back of the bow, in order to achieve v1. These light barriers delivered a start and stop signal when the bolt passed.

Maximum speeds were 53 m/sec for crossbow 1, 61 m/sec for crossbow 2.

Bolt speeds ranged, depending on bolt weight, from 39 to 50 m/sec for crossbow 1, and from 46 to 55 m/sec for crossbow 2.

Remarkable was that the speed of both crossbows decreased slightly after some shots and settled in an area of approx. 11 to 14 % below maximum speed.


4 Summary 

Test scenario observed a number of 43 shots (using different reconstructed medieval bolts), fired from two reconstructed medieval crossbows onto three different targets.

Target 1, as well as the core of target 2 and 3, was set up by blocks of ballistic soap. Ballistic soap is similar to gelatine. Both are used as a material for simulating human body, yet experiments have shown that gelatine generally allows deeper penetration, partly twice as deep as soap does. Beside speed measurements, penetration depths in the targets as well as the perceived damages in the protective arming are documented.  Bolt speeds reached maximum speeds of 53 m/sec (crossbow 1) and 61 m/sec (crossbow 2).

Penetration depths  with a distance of 10 m for target 1 lie between 143 and 208 mm, in the ballistic soap for target 2 between 49 and 91 mm and for target 3 between 46 and 66 mm (referring to bolt 1).  The heavy bolt 4 was not able to penetrate the protection (target 2 and 3) and add to the bearer maximum a bruise.

Tests also showed that authentically dimensioned textile armour can give effective protection against shots from crossbows, especially against bolts equipped with heavy, blunt pyramidal heads. If textile armour is additionally protected by mail, protection value raises by 5 up to 27 % (whereas the “value” of mail depends on ring strength, ring density). 

Nevertheless it has to be said that a higher protective effect depends on factors like mass (weight) or mobiliy. However, all-embracing protection was given at no time, since the already mentioned race between protective and attacking weapons is a phenomenon going on until the present day.


5 Literature 

Harmuth Egon, Die Armbrust, Akad. Dr.- u. Verl.-Anst., Graz 1975. 

Payne-Gallwey, The Book of the Crossbow, Dover Publications Inc., New York 1995. 

Sudhues Hubert, Wundballistik bei Pfeilverletzungen, Münster 2004. 

Zimmermann Bernd, Mittelalterliche Geschossspitzen, Kulturhistorische, archäologische und archäometallurgische Untersuchungen,  Schweizerischer Burgenverein (Hrsg.), Schweizer Beiträge zur Kulturgeschichte und Archäologie des Mittelalters, Bd.26, Basel 2000.


[1] Payne-Gallwey 1995, p. 62.

[2] Payne-Gallwey 1995, p. 22.

[3] Harmuth 1975, p. 60.

[4] Bow strength measured immediately after stringing the bow. Bow strength depends on organic materials as well as on variable surrounding temperature: higher temperatures lead to a “softer”, lower temperatures to a “harder” bow.

[5]See note 4

[6] Zimmermann 2000, p. 81f.

[7] Zimmermann 2000, p. 72.

[8] Sudhues 2004, p. 125.