How Heavy Was the Shield from the Fayum?

In May a well-known ancient historian told the Internet that the style of shield from Kasr el Harit in the Fayum in Egypt weighed 10 kg. He probably got this from Mike Bishop and John Coulston’s classic handbook Roman Military Equipment (second edition p. 62) which cites reconstructions by Peter Connolly and Marcus Junkelmann. He was writing a general lecture so could not spend too much time questioning his sources. But I think this estimate is too high for four reasons.

First is a back-of-the-envelope calculation. Most of the information on this shield comes from an article by Wolfgang Kimmig who examined it in Cairo. He says it was 128 cm long and 63.5 cm wide but does not explain whether he measured the width along the curve or across the hollow of the shield. Assuming that he meant the later, lets pretend that this shield was a rectangle 128 x 70 cm (ignoring that the corners are rounded off). Kimmig said that the individual strips of wood were 2 to 3 mm thick at the rim and somewhat thicker in the center. He spoke of “a thickness of less than 1 cm measured at the rim” with three layers of 2-3 mm thick planks and “somewhat greater in the center” but let’s assume an average thickness of 1 cm of wood which is higher than many other European shields. Then the volume of the shield board is 128 cm long x 70 cm wide x 1 cm thick = 8690 cm³

The only other parts of the Kasr el Harit shield were the wooden boss and the spine along the long axis. Using the good old formula V = 4/3πr³ the volume of a spherical shell 6 cm in diameter and 1 cm thick is 381 cm³. A dowel 1 cm by 1 cm by 128 cm is another 128 cm³. So lets estimate the volume of the wood as 9469 cm³.

Kimmig was not sure of the wood but thought it might be birch. Birch averages around specific gravity 0.5 to 0.7 depending on species and moisture content. So the wooden parts of the shield should weigh about 5681 g (that is too many significant figures but lets round off at the end of the calculation).

There were no large metal parts on this shield such as reinforcements for the boss or the rim. If there had been, some of the wooden parts might have been smaller or thinner. So lets ignore those.

This shield was covered on both sides with felt, but Polybius’ shields were covered on one side with linen and rawhide. If we assume that is ‘canvas weight’ linen at around 300 g / m2 that adds 0.8960 × 300 = 268.8 g for the linen. Rawhide is less dense than water (as anyone who has tried to untangle a dog chew by soaking it knows). I measured some of the roughly 1 mm thick water buffalo rawhide I am using for my shield projects and got a density per area of about 0.09 g / cm². A random source on the Internet gives density per volume of 0.86 g / cm³ for “leather, dry” which would give a similar estimate. (Twenty years ago I got numbers like that from “Weights of Common Substances” by Andrew Roy which was hosted at www.tesarta.com/www/resources/library/weights.html and now at http://www.mythosa.net/).

So one layer of canvas and one layer of rawhide on one side of the shield would add about 1075 g.

So a back-of-the-envelope calculation suggests that a shield the size and shape of the Kasr el Harit shield should weigh about 7 kg if its covered on one side with canvas and rawhide (5.681 kg of wood + 1.075 kg linen and rawhide rounded to one significant figure). That is noticeably less than 10 kg, even though we tried to make our estimate as high as possible.

Second, as I show in my article “Get to the Point!” the first reconstructions of ancient artifacts are usually 50-100% overweight. Our replica Roman shields are usually made from plywood and the same thickness everywhere, but the Kasr el Harit shield was thinner at the edges and thicker in the centre. In the 1970s nobody knew what weight of leather to use, but we now know that must surviving shields from later times are covered in thin skin like parchment. We are also tempted to add lots of metal bits to our shields, because they show up in archaeological reports and look shiny. We tend to make them of thicker metal than the originals because to us metal is cheap, because we don’t have to march with them for months on end, and because the originals are fragile. The Kasr el Harit shield had no iron or bronze parts larger than a nail. The 1970s and 1980s were a pioneering time in ancient reenactment, not long after John Coles dressing up in tweeds to hack at the edge of a thin copper shield and announce that he had proved that thicker bronze shields used by skilled warriors were useless. Reconstructions from that period deserve to be reconsidered.

Third is that Wolfgang Kimmig (German Wikipedia), who seems to be the last person to handle the shield and publish what he saw, described it as “astonishingly light.” 1940 was a different time but even in 1940 professors were not ironworkers or longshoremen.

Fourth is a quick-and-dirty reconstruction by Matthew Amt sometime before 2003. His copy of this shield came out at “less than 15 pounds” (7 kg) even though he expected it would be heavy because it was so big. Edit: in March 2024, Sergio Escudero of Ars Scutae in Tudela, Spain copied the Fayum shield in hand-made poplar plywood and wool felt. His replica weighs 4.8 kg (no link because its Facebook which does not have stable, login-free links).

The Kasr el Harit shield is quite big at 128 cm long. I expect that Roman shields were heavy for infantry shields, but a “heavy” shield seems to have been around 6 to 7 kg not 10 kg (good replicas of Viking shields, with similar surface area, weight 3 to 4 kg, smaller Migration Era shields can be even lighter). I don’t have time and resources to make and document a reproduction this year, but I suspect that when someone does, they will find a weight in the 5-7 kg range (follow https://patreon.com/dimicator for updates on a project at the Varus Battle Museum in Germany which is reproducing Imperial Roman shields). It would have been very easy for Connolly and Junkelmann to use very heavy leather on both sides, or not taper the wood to be thin at the edges and thick in the middle, and unconsciously add weight. Science is all about reevaluating what we think we know as our understanding improves and new evidence becomes available.

Because it has no detailed archaeological context and has never been scientifically dated, we don’t know when this shield was made or whether it was La Tène Celtic, German, Roman, or Egyptian in the Celtic, German, or Roman style. We don’t know some of its dimensions. There are so many questions. But given what we know about ancient shields and modern replicas today, I suspect that the original weighed more like 6 or 7 kg than 10 kg.

Further Reading

Kimmig, Wolfgang (1940) “Ein Keltenschild aus Ägypten.” Germania Bd. 24 Nr. 2, pp. 106-111 http://nbn-resolving.de/urn:nbn:de:bsz:16-ger-413717

Junkelmann, Marcus (1986) Die Legionen des Augustus: Der römische Soldat im archäologischen Experiment. Kulturgeschichte der antiken Welt, Band 33 (Philipp von Zabern: Mainz am Rhein) pp. 174-176 {reports a weight of “no less than” 9.65 kg without the leather case and carrying strap

“Get to the Point: What Questions Should We Ask About a Spear?” in Graham Wrightson (ed.), Proceedings of the 2022 International Ancient Warfare Conference (Cambridge Scholars Press) in press as of April 2023

A third edition of Bishop and Coulston is in preparation https://romanmilitaryequipment.co.uk/roman-military-equipment-edition-3

Uta Schröder, Ingo Petri, Katrin Zerbe, “Computertomographische Messungen an den frühkaiserzeitlichen römischen Schilden aus Kalk­riese (Lkr. Osnabrück). Neue Erkenntnisse zu Aufbau, Herstellung und Vorteilen römischer Sperrholzschilde,” Archäologisches Korrespondenzblatt, Jahrgang 52 Heft 1 (2022) pp. 91-107 https://journals.ub.uni-heidelberg.de/index.php/ak/article/view/94325 discusses the thickness of the edges of shields from the early Roman empire. See especially p. 105 n. 47 “Die Abbildungen bei Junkelmann 2015, Abb. 259 und 2008, Abb. 165. 251. 272. 287. 337. 364-365 lassen erkennen, dass die Schilde ihre Stärke auch bis zum Schildrand beibehalten.” They point to reconstructions of the Kasr el Harit shield which weigh 5.28 to 7.84 kg and suspect that the lower end of that range is more correct for the Augustan period.

Rolf Fabricius Warming, René Larsen, Dorte V. P. Sommer, Luise Ørsted Brandt, and Xenia Pauli Jensen, “Shields and hide – On the use of hide in Germanic shields of the Iron Age and Viking Age,” Bericht der Römisch-Germanischen Kommission Band 97 (2016) pp. 155-225 https://www.academia.edu/44259256/Shields_and_hide_On_the_use_of_hide_in_Germanic_shields_of_the_Iron_Age_and_Viking_Age

John Coles, Archaeology by Experiment (1973) is supposed to have the poorly-designed experiment with shields but I last read it 15 years ago so check the reference before citing!

Buckland, Paul (1978). “A First-Century Shield from Doncaster, Yorkshire.” Britannia, Vol. 9 pp. 247–269 https://doi.org/10.2307/525941 The poor chaps built their replica of this auxillary shield to weigh 20 pounds / 9 kg and thought it was underweight! (page 259)

Edit 2023-11-18: Our library decided that its copy of Junkelmann’s Legionen des Augustus was really lost, so I ordered a copy by interlibrary loan and checked the reference. Junkelmann p. 176 mentions a “Berlin relief” where a praetorian seems to hold his shield with just one finger through the shoulder strap (does he mean this relief?)

Edit 2023-11-21: hi birdsite! https://nitter.net/EquipmentRoman/status/1726544663983644712#m

Edit 2023-11-22: Mike Bishop informs me that Junkelmann’s “Berlin relief” was probably University of Pennsylvania Museum of Archaeology and Anthropology, object number MS4916A https://www.penn.museum/collections/object/115536 (acquired in 1910). For more see E. H. Hall, “A Roman Relief from Pozzuoli,” The Museum Journal, Volume IV, Number 4 (December 1913) pp. 142-146 https://www.penn.museum/sites/journal/294/

Edit 2023-11-24: added a link to Schröder et al. 2022

Edit 2024-01-02: Eckardt, Hella, and Walton, Philippa (2021) Bridge over troubled water: The Roman finds from the River Tees at Piercebridge in context. Malet Street: Society for the Promotion of Roman Studies. https://doi.org/10.5284/1085344 p. 88 n. 37 “12 examples of (copper-alloy shield) binding from the Piercebridge excavation were catalogued but were thought too narrow to enclose the thickness of a shield (Allason-Jones 2008, 11.65).” Thanks Mark Shier.

Edit 2024-03-27: as a placeholder see the weights of four Marburg shields https://www.patreon.com/posts/basic-stats-of-36465569

(scheduled 17 June 2023)