On Trip-Hammers and Rolling Mills

Over on his site Bret Devereaux has a good description of a problem in Roman Military Equipment Studies. In book 6 of his histories, Polybius says that most Roman infantry wear a plate of bronze a span broad called a kardiophylax “heart-protector” on their breast, except for the wealthy who wear coats of mail. No such plate survives from a Roman site after 300 BCE, and no sculpture or painting shows it. As Roman rule expands across Italy, locals stop building tombs with detailed paintings full of arms and armour, and body armour tends to be a rare find. By the fifth century BCE, Samnites and Campanians had replaced simple disc breastplates with more complex arrangements of a breastplate, a backplate, a bronze belt and armoured straps over the shoulders and under the arms. We therefore have to assume that Romans either reverted to a style of armour from several hundred years before, or that Polybius’ description just mentions one-part of a seven-part armour. To my knowledge, no other surviving writer says that Romans wore such a breastplate, and there are no carvings or paintings which show Romans wearing them (Varro’s pectorale was made of strips of leather, De Lingua Latina 5.24). Both interpretations match objects from the ancient Mediterrean, and both match later armour from other cultures such as the “good iron for his body” worn by Robert the Bruce’s militia in 1318¹ and char-aina “four mirrors” armour in the Persianate world. I am doubtful that most Romans could afford not only a helmet, a sword, and and iron-bound shield but most of a bronze breastplate, but Devereaux is more confident. There are a lot of things to think about here, such as why the Roman Republic, a relatively egalitarian society, did not leave much art which showed ordinary soldiers. However, this week I will write down my thoughts about one technical question which I took the time to work through.

I think that if you wanted to make a bronze breastplate in the ancient world, you took an ingot and hammered it thin (possibly starting by melting it and pouring it onto a broad flat stone). You need to stop frequently, heat, and cool in air or quench in water to anneal it so it does not get too hard and brittle and crack. Devereaux thinks that they were “probably produced from sheet metal (sheet bronze, in particular), rather than forged from an ingot.” In a comment he referred me to David Sim in the UK.

I don’t own Sim’s book Iron for the Eagles but I do own Roman Imperial Armour (Oxbow Books, 2012) by David Sim and J. Kaminski which presents Sim’s theory (and Sim’s article in Journal of Roman Military Equipment Studies 2003/4). Sim’s argument seems to take the following form (pp. 49-57): some armour from the Roman empire has smooth, parallel surfaces and consistent thickness. I don’t know any way to do this by hand or with powered up-and-down motion, but I can do it with a pair of rollers turned by hand and an assistant feeding metal between them. Such rollers would likely have been recycled when no longer needed and mounted in wooden machines, therefore would have left no archaeological evidence.

I don’t know how common the iron or copper-alloy sheet with smooth, parallel surfaces actually is. Table 7 of the book shows that one part of a Roman armour or shield usually varies in thickness across its surface by a factor or 2 or 3 which is typical of low-tech armour. The main examples of smooth parallel surfaces seem to be scale armour and layers within a sheet of iron or steel, and the paper on that steel notes that the same could probably be achieved with a trip-hammer (Fulford, Sim, and Doig 2004: 201). A trip-hammer is a long hammer on a pivot, whose long end comes close to a turning shaft driven by water or a motor. Pegs on the shaft lift the hammer and cause it to strike an anvil in a steady rhythm with more than human force, so the operator just has to focus on positioning the work.

Whereas trip-hammers were widely used in the Roman empire, late medieval Europe, and China, roller mills for sheet metal only appear in Europe in the 18th century, and mills for making bar iron into rods for nails only in the 16th century. That is almost 2000 years later than Polybius. Forming two cylinders and mounting them parallel to one another with handmade tools is harder than it looks, and making an adjustable mill that can produce more than one thickness is even harder. It was hard to justify specialized machines before the 20th century, because labour was cheap and markets were unstable. You could not rely on making one specialized thing for a distant market, because one day the seas would be closed by pirates or the road over the mountains would be shut to keep the plague out. If you wanted to keep in business, you needed to produce a variety of goods, or simple and versatile goods like nails. Copper alloys were traded in ingots, and iron was traded in bars, because skilled metalworkers could make anything you wanted from the raw material. If you bought specialized pre-made forms, you might find that there was no market for finishing them.

To make a breastplate, you do not want sheet metal of uniform thickness. You will want the front and center to be thicker and stronger, and the sides to be thinner, because you are most concerned with being stabbed or shot through the vital organs by someone in front of you. As you stretch the metal to fit around your body, it will get thicker in some places and thinner in others, and if you have time to take care you will put the thick metal in the middle and thin near the edges. Nobody but you and your servant will ever see the inside, so it does not matter if the inside is lumpy and oxidized as long as the outside is smooth and beautiful. So there would be no benefit to using rollers, as there would be for making 0.3 mm stock for scale armour and applied decorations.

Sim’s approach in Roman Imperial Armour seems quite a contemporary one, with a focus on specialized labour-saving machines, uniform properties and dimensions. Moving between contemporary and traditional approaches often takes some adjusting, as when I volunteer on carpentry projects with no wood-to-wood joinery and everything butted and nailed together and my inner fourteenth-century person has to take a lie down and drink some ale. Reenactor Matthew Amt has some disagreements with how Sim makes pilum heads and spearheads, and a specialist in reproducing ancient metalwork did not like the book:

Roman Imperial Armour: The Production of Early Imperial Military Armour overreaches in basing so many of its assertions on the limited experience of a hobbyist blacksmith. It would be presumptuous for a metal-detectorist to write a treatise on archaeological practice, but for some reason academics think that any experience at all qualifies them to extrapolate on ancient craft industries.

no citation because this was from a private forum discussion! Ask me in person if you want to know who said this.

Iron for the Eagles has a very good reputation and I hope to read it one day.

Also, a roller mill is a crank-based mechanism. Cranks were probably invented in Iberia around the fifth century BCE. In the last few centuries BCE, Italians were starting to adopt them for purposes like milling grain. The famous rotary querns from Pompeii were a new and high-tech way of making your flour. It is only in the past few decades that it became clear when and where rotating millstones were invented. Sim and Kaminski know some of this history because they cite L. Sprague de Camp’s 1963 The Ancient Engineers. However, they don’t notice that if cranks were a new technology, its unlikely that they were immediately applied to milling. Europeans had been using cranked grindstones for at least 700 years before they started to use cranked roller mills for metalworking. That suggests that adapting the crank mechanism to metalworking is not an obvious next step. And central Italy in the third and second century BCE was not a center of mechanical and metallurgical innovation, but a place which was dazzled by Celtic swords and Hellenistic devices like the Antikythera mechanism. Slaves from the east seem to have replaced Italy’s native textile traditions by the end of the Republic because they were cheap and clever and it was easier to let them do what they knew how to do than retrain them.

There were a few ares where Imperial Roman technology was similar to 16th-18th century Europe, such as the size of ships and the availability of iron. Celts and Romans did use a lot of thin iron, bronze, silver, and brass sheet, and making such sheet by hand and then smoothing it out is time-consuming and wasteful. The shield from Dürrnberg grave 373 had many iron reinforcements just 0.15-0.20 mm thick after corrosion.² I am open to the possibility that trip-hammers or stamping mills were used to beat out sheet in the Roman empire, since they were used for other tasks. I might even be convinced that rolling mills were used in some places for specific purposes like making very thin brass sheet for strapends, lanterns, and scale armour. The Roman empire had large-scale production in centralized locations for distribution by sea and fairly extensive use of water power. However, I would want much more evidence to believe in roller mills in Polybius’ Italy, and I expect that if some shops had trip-hammers to bang out helmets and buckets, others just used teams of sweaty men with hammers. As late as the First World War, some firearms were made in giant factories, and others were made with hammers and files in one-room workshops in Spain.

There are lots of threads I would like to pull further, like when sheet metal in standard gauges first appeared, and my onetime teacher John Peter Oleson’s argument about whether the ships from Lake Nemi in Italy had cranks on their pumps (in 1984, he thought the cranks were “archaeologically a fantasy” based on misinterpreting fragments of wood). I never got around to finishing Panagiota Manti‘s thesis on casting ancient Greek helmets. However, I hope that this shows why I am open to the use of articulated cuirasses in Polybius’ Italy, somewhat less open to the use of trip-hammers, and quite skeptical of the use of roller mills or armourers who started with sheet metal. Ancient metalworking is an exciting topic where archaeologists, metallurgists, and makers can work together, but its hard to keep up with.

Unlike ancient metalworkers, I am not secretive about my methods! But unlike many ancient writers I am not independently wealthy. If you can, please support this site.

Further Reading

Michael Fulford, David Sim, Alistair Doig, “The production of Roman ferrous armour: a metallographic survey of material from Britain, Denmark and Germany, and its implications,” Journal of Roman Archaeology, Volume 17 (2004) pp. 197-220

(scheduled 6 April 2026)

Edit 2026-04-06: first photo is in the wrong museum! Had a quick look at the book by Oleson.

1. https://www.rps.ac.uk/mss/1318/29 I will discuss and translate this in my second article on linen armour for Medieval Clothing & Textiles ↩︎
2. https://doi.org/10.11588/jrgzm.2009.1.16569 I don’t yet have a Zotero plugin in this browser to paste bibliographic details, but you can find them in “Plywood shields in European history,” Journal of Roman Military Equipment Studies 23/24 (2022/23) pp. 9-23 ↩︎