How Much Does the Protection of Low-Tech Armour Vary?

Designers of roleplaying games who are interested in learning how the real world works, and not just studying other people’s stories and games, usually put a lot of thought into the combat mechanics. One old argument is about how to handle the performance of armour. Fairly early on (sometime in the 1970s or 1980s?), the idea of a damage roll was combined with the idea that armour could provide a penalty to damage. However, this tends to bother people whose archetypical combat involves modern firearms and armoured vehicles or kevlar body armour.

Bullets and shells have a very predictable ability to penetrate armour, and modern industrialized, standardized-tested armour has a very predictable ability to resist it, and the damage-roll-minus-armour model tends to let some things get through which should be stopped. While sometimes this can be abstracted away (“eh, maybe those few points of damage represent bruising”) other times that is difficult to justify (“did the shell explode inside the tank or outside? Did the Deathly Dagger of Draining touch his flesh or not?”) One solution to this is to treat both penetration and resistance as more or less fixed, then generate the effect of the wound based on their interaction. GURPS fans often refer to this as armour-as-dice, because armour can be treated as reducing the predictable number of damage dice which the attacker rolls instead of the variable results of that roll.

However, models which treat penetration and resistance to penetration as more-or-less fixed tend to make people who are more interested in combat with hand weapons uncomfortable. In this post, I would like to explore what we know about how much the ability of hand-made armour to resist weapons can vary, even within a given piece of a known form and quality. If you want, you can skip to where I sum up.

One obvious issue is thickness. Traditional armourers had no reason to make every part of an armour equally strong, and very good reasons not to. Some parts of the body are more exposed to powerful blows, and some parts are more sensitive to injury. Given a limited budget of mass, armourers saved weight on parts which were unlikely to take strong blows or would not be life-threatening if wounded, and redistributed it to places which needed more protection. Without tools like rolling mills, there was no disadvantage in this: turning a flat piece of steel into a helmet which is strong front-and-centre and weak in the back is no harder than turning it into a helmet which is more or less equally thick all over.

In a lecture Toby Capwell cites the visor of an armet in the Wallace Collection (number A152?) as being about 6 mm thick front and centre and 2 mm near the hinges at the sides, and that strikes me as more variable than average but not remarkable. An early Corinthian helmet in Manchester Museum varies in thickness from 0.08 to 0.20 cm depending where it is measured (for example, thickness along the bottom of the neck-guard varies from 0.08 to 0.18 cm). Soft armour often had less variation, but John Howard the Earl of Norfolk once ordered a doublet of defence with nine layers of cloth in the sleeves and 23 in the front quarters, and Tasha Kelly measured the padded armour stuffed with cotton in Chartres Cathedral as varying between 0.5″ and 1″ thick depending on whether fashion demanded that an area be slender or pluffy. The difference in thickness between different parts (eg. a breastplate and a backplate) can be even greater, and many parts of the body are protected in some areas by several overlapping pieces and in others by only one. The yoke and shoulder flaps of a tube-and-yoke armour overlap the body, and those armours sometimes overlapped at the vertical opening in the body. So thickness can easily vary by a factor of two or three across a single piece. Using a formula by Blyth and Atkins, this increases the energy required to penetrate it by a factor of 3.0314 to 5.7995 (so say 3 to 6 after rounding to one significant figure). Some people don’t agree with Blyth and Atkins that the energy required to penetrate plate by stabbing is proportional to the thickness to the 1.6th or 2nd power, but I don’t know that any of them have published alternative formulas in a peer-reviewed venue.

Pieces of armour often incorporate different materials. Scale armours often mix rawhide (for cheapness and lightness) and metal (for strength). These can be alternated in a given row, or the stronger materials can be concentrated in the areas which need more protection. Another solution is armours which combine large plates with mail, such as the front half of a cuirass in Milan. The ancient Greeks famously reinforced some parts of their soft armour with scales (bronze? rawhide? iron?) and left others bare.

The quality of materials can also vary inside a given piece. Before the 20th century, “iron” and “steel” often contained a significant amount of slag unevenly distributed through the metal. It was common to enrich the surface of armour with carbon, creating armour which combined hardness and flexibility. Forging, earlier damage, and failures in heat treatment or case hardening could further weaken some spots and strengthen others. Mail and textile armours tend to have less variation, but stuffing can vary in quality as well as thickness. I don’t know of any tests against plate armour made from similar materials to different grades of historical armour. A few tests destroy original armour, but these usually pick a handful of cheap pieces rather than making holes in a wide range including ones of good metal. In The Knight and the Blast Furnace, Alan Williams waved his hands and estimated that medium-carbon steel without heat treatment (*** metal) might be 2.2 times as effective for a given thickness than slaggy iron (* metal). He did not have much evidence, but his work is probably good enough for gaming. I don’t know of any data about how much the quality of early plate armour could vary within a piece, but it might be on the same order.

Armour also has complex shapes curved in three dimensions. A good rule of thumb is that no point on a piece of plate armour is flat. Because of these shapes, the angle of impact and effective thickness can vary widely (especially against plate armour, the kind whose properties are least poorly understood). Velocities are low enough that the motion of attacker and target, or the presence of horses which increase the speed of impact, can also make a difference. Plate armour usually has rounded, smooth, hard surfaces designed to direct oncoming weapons away from the body. I won’t try to quantify these things, just say that it is common to strike at a bad angle where the weapon fails to bite.

Summing Up

• So depending on where you hit a piece of armour, the thickness can vary by a factor of 2:1 or 3:1 (and the energy required to penetrate by a factor of 3 to 6) …
• materials can vary as widely as 3 mm rawhide and 1 mm medium-carbon steel within a single piece …
• the quality of materials can vary the energy required to penetrate plate armour by at least a factor of 2.2 …
• and you can do your own trigonometry to explore how the angle of impact affects strength

Less formally, we know that armour was often rated as proof against particular weapons, and sometimes tested against them. Sources from both Europe and Japan describe this, and no doubt warriors everywhere have taken a swing at some new armour. Warriors in many cultures were comfortable teaching rules of thumb like “stab at the gaps in the plate.” Many of these rules seem reliable, but whether they were true 90% of the time or 99% is another question. There is a beautiful passage in a 15th century text which assures readers that the author has hardly ever seen someone wearing a certain style of armour be killed “at least if they were men accustomed to fighting” and soldiers everywhere have a healthy respect for the gods, Fortune, or Friktion. Where the designers of modern bullet-resistant vests seek certainty, most traditional armour seems to have been designed to stop some common threats most of the time as long as nothing went too badly wrong.

There are plenty of reasons why someone designing a roleplaying game might not decided to try to accurately model the performance of low-tech armour. Many stories focus on characters with superhuman ability to defeat armour. Sometimes this was because the person telling the story overestimates what human strength can do, and other times it is a deliberate way of increasing the drama or impressing the audience. Many stories brush over many ineffective attacks to focus on the few decisive ones. Both the Iliad and storytellers in New Guinea use this trope. We don’t know very much about the performance of any kind of low-tech armour other than plate, or how different factors influence its performance, so its hard to show that one model is less wrong than another. And of course, roleplaying games don’t have to focus on combat! However, I hope that this post gives food for thought to people who are interested in knowing how much the ability of low-tech armour to resist muscle-powered weapons varies, and helps people who are more familiar with modern firearms understand why experts in earlier weapons don’t share their outrage with armour which sometimes stops an attack and other times fails against a similar one.

Further Reading:

• P.H. Blyth and A.G. Atkins, “Stabbing of metal sheets by a triangular knife. An archaeological investigation,” International Journal of Impact Engineering 27 (2002) pp. 459-473 {look at formulas (2) and (12) and page 469 … Blyth’s PhD thesis on Greek armour and shields uk.bl.ethos.450089 is a classic} ↑
• A. H. Jackson, “An Early Corinthian Helmet in the Manchester Museum,” The Annual of the British School at Athens, Vol. 99 (2004), pp. 273-282 http://www.jstor.org/stable/30071538 ↑
• Tasha D. Kelly, “The Tailoring of the Pourpoint of King Charles VI of France Revealed,” Waffen- und Kostümkunde Heft 2/2013 p. 159 http://cottesimple.com/pourpoint-of-charles-vi/pourpoint-article-available/ ↑
• Sylvia Leever, “For Show or Safety?” Arms & Armour 3.2 (2006) pp. 117-125 {contains a test of bullets against 17th-century breastplates and a formula for penetration of homogeneous iron plate relative to energy and bullet diameter; the book by Williams below and a study of the armoury at Graz are other studies which destroy original armour} ↑
• Robert W. Reed, Jr., “Armour Purchases and Lists from the Howard Household Books,” The Journal of the Mail Research Society 1.1 (July 2003) pp. 25-38 http://www.erikds.com/pdf/tmrs-journal-1.pdf {modernizes the original Middle English, I have no problem with his rendering of the description of the layers} ↑
• Alan Williams, The Knight and the Blast Furnace (Brill) ISBN-13 9789004124981 {tries very hard to draw clear and simple conclusions from complex, contradictory, gappy evidence; he has been trying to get his publisher to reprint it at a more reasonable price for years} ↑

An acquaintance with a PhD in physics from a university in the Pacific Northwest has published a critique of the formulas used in GURPS as “Under the Hood: The physics of projectile ballistics” http://panoptesv.com/RPGs/Equipment/Weapons/Projectile_physics.php His article lacks citations and explanations of terms, I have trouble reading his formulas which are written as text strings without markup, and since I am not an engineer I prefer to rely on formal publications which have been approved by experts in the field rather than trusting my ability to judge the math and the logic. However, readers with a background in engineering or ballistics might want to take a look. It is certainly possible that the few engineers and metallurgists interested in the performance of plate armour have missed a body of research by engineers in other fields.

• Edit 2017-01-07: Fixed typo in Blyth and Atkins note and added link to Blyth’s thesis.
• Edit 2018-09-23: Added the link to the ordinance of Louis XI
• Edit 2020-10-23: Vincent le Chevalier has estimated the energy of a cut with a long, straight, one-handed sword by using a video and measurements of the weapon and applying some intermediate kinematics. He estimates a peak transverse kinetic energy as the sword reaches the level of the opponent’s face at around 23 J. For comparison, Horsfall et al. (Forensic Science International 1999) found the mean energy of an underarm stab by male students with a small knife to be 28 J (Connolly et al., Journal of Battlefield Technology 2001 had a lower figure for spear thrusts but they only did 3 or 4 tests per stabbing method).
• Edit 2021-08-30: Converted to block editor, fixed an image link Wade Allen has some comments on the variation in thickness in 16th and 17th century European burgonets:

I do have similarly formed burgonets. We took my nice new Acme Armour Caliper (John delivered it) to a couple of them. The ones we played with are munition pieces so they are pretty thin. A general characterization would be about .05 in. (1.3 mm – ed.) in the skull, but there is a lot of variation. I found .028 (0.7 mm), and I found some spots up to .07 (1.8 mm). One of them is closer to .040 (1.0 mm). These aren’t really designed to stop everything, but they would help keep you from being killed a few times.