This week, we close out our four(and a half)-part (I, II, III, IVa, IVb) look at pre-modern iron and steel production, although I ought to note that there will be at least one addendum discussing pre-modern cast iron and crucible steel (Wootz) production. Last week, we looked at the processes used to create steel from iron by introducing carbon (rather than the modern method of producing steel from pig iron by removing carbon). That process changes the characteristics of the metal, since steel is harder and more elastic (bending and springing back, rather than bending and staying bent) than iron.
But the chemical composition of the iron or steel isn’t the only factor in determining its hardness and ductility: mechanical processes and heat treatment also alter the internal structure of the metal. Since the pre-modern blacksmith works with both of these – mechanical processes through hammer and heat through the ‘heats’ of forging – he cannot afford to disregard the changes that both bring. As we noted in part III, the blacksmith’s job is not merely to bring the metal into the necessary shape for its final application, he also has to get it there with the right characteristics – the correct balance of hardness, strength, ductility and elasticity to get the job done, whatever that job might be.
And I should note again that different tools – and indeed, different parts of the same tool – might have very different demands in this regard. Metal files needed to be prodigiously hard so that they could abrade already quite hard materials. The tips of chisels and picks also needed to be very hard, to keep their shape under repeated impacts, but not so hard that they broke; some ductility to absorb the energy of impact was needed. Armor needed typically to only be somewhat hardened (though demands of this sort get higher with gunpowder) and can be fairly malleable; the armor plate that deforms elastically on impact still absorbs the strike. Swords made varied demands: their significant length demanded strong steel with good elasticity to be able to spring back into proper shape after absorbing the forces of swings and impacts, while keeping a good cutting edge demanded a high degree of hardness (hardness being the characteristic that determines how sharp an edge can be and also how well that sharpness can be held during use).
So a blacksmith couldn’t simply maximize one trait at the expense of the others. Nor could they have just one kind of steel, with just one set of characteristics for every application. Rather the learned techniques – copied and practiced during their apprenticeship – were developed, probably over generations, by experimentation to produce tools, weapons and armor each with its own unique ‘right blend’ of characteristics, though two major methods we will discuss today (along with, of course, carbon content, which we discussed last week).
As always, if you like what you are reading here, please share it; if you really like it, you can support me on Patreon. And if you want updates whenever a new post appears, you can click below for email updates or follow me on twitter (@BretDevereaux) for updates as to new posts as well as my occasional ancient history, foreign policy or military history musings.
Work hardening refers to the tendency of metals, when hammered (or strained into any kind of plastic deformation) to become harder. In its pristine state, the atoms of a metal are generally arranged in a crystal structure, joined by metallic bonds. These lattices of atoms are neat and regular. When that neat, regular lattice of atoms is strained, resulting in plastic (rather than elastic) deformation, those neat regular lattices get scrunched and bent and otherwise dislocated. This happens in iron, but also in most metals – indeed, work hardening is much more important in bronze-work than it is in iron-work, though it cannot be neglected in either.
That dislocation makes the metal resistant to further deformation (plastic or elastic), both increasing the hardness of the metal and its yield strength (the amount of energy you need to apply for plastic, rather than elastic deformation to take place), but at the same time makes the metal less malleable and more brittle (that is, more likely to break than bend). This is a property that must have been apparent to the earliest copper-smiths: as they hammered their copper into shape, each blow made the metal slightly more resistant to hammering, until eventually it would refuse to budge almost completely (and possibly break instead). But the same process works for iron, though because of the different production process for iron, it has to be a touch more intentional.
It is possible to reset this process through a process called annealing. When a work hardened metal is heated up, the energy provided by the heat allows the atoms to break and reform their bonds, causing the crystal structure to return to its normal lattice and removing the various dislocations (while retaining its new shape). To completely anneal iron, it is heated up to its annealing temperature (which varies based on the carbon content, as you can see in the chart below) and held there for an extended period, which will cause it to get a soft as possible for the given iron and carbon content. This can be very handy if the iron needs to be cold worked (see below).
If the iron is not held at the high temperature, but is allowed to cool slowly in the air, it is said to be normalized (this process, normalization, is a subset of annealing). As with full annealing, the heating process works out all of the little strains in metal (heats for this process range from 700 to 900°C), but by cooling it slowly afterwards (rather than holding it at a high temperature), it causes the actual size of the atomic lattice of the metal to shrink (the technical term is ‘grain refinement’ as the individual grains of the lattice are ‘refined’ – meaning ‘get smaller’), which results in a metal somewhat harder than in a fully annealed state. Normalization is an important step for metal that is going to be subsequently heat treated as well.
Now the attentive reader will be thinking, “but wait, if heating the metal up to forging heat and letting it cool in the air resets the work hardening process – well, we’re doing that every time we heat the metal for forging.” Indeed! Hot working (which is what we’ve been describing so far with forging) generally does not meaningfully work harden metal for this very reason. And for a smith working with good steel that is going to be heat treated (which we’ll discuss in a minute) this doesn’t matter a whole lot.
But – for reasons which we’ll discuss – iron with very low carbon contents cannot be heat treated for hardness. For a smith working with nearly pure (no carbon) iron, work hardening is his best option for increasing hardness and yield strength. In this case, the iron object would typically be brought essentially to its final shape and then finished by being hammered cold (‘cold working‘). While copper and bronze are generally soft enough to be almost entirely cold worked, iron really isn’t, except in fairly thin sheets, so the bulk of the actual shaping will still be done with hot forging, but with a final phase of cold hammering which will induce work hardening (and also, incidentally, can remove tool marks from the hot working, which might be cosmetically desirable).
Heat Treatment Basics
The changes steel undergoes when it is heated and cools can also alter its characteristics. Because, as we’ll see, a big part of the effect of heat treatment has to do with the carbon atoms diffused within the crystalline iron structure of steel, pure iron and low carbon steel cannot be effectively heat treated and has to instead be work hardened in order to achieve similar results. Heat treatment allows the blacksmith fairly fine control over the hardness, strength and ductility of the steel and can even allow for different parts of a single steel object to be hardened to different degrees.
For high-carbon steel, the heat treatment cycle is often called ‘tempering and quenching’ although as we’ll see it would perhaps be more accurate to call the process hardening, quenching and tempering (and then quenching one more time) to get the correct order. It is something of an irony that first, intense quenching of raw, red-hot steel (the second quenching is much less dramatic as it occurs at much lower temperatures) – almost always shown as the final step in blacksmithing in film or video games – is essentially never the final step, for reasons that will soon become apparent.
First, some complicated metallurgy and then we’ll get to the actual real world processes a blacksmith would use. High carbon steel is initially a mix of two kinds of iron, ferrite and cementite, the former being a cubic atomic structure of pure iron, whereas the later is an iron-carbide (Fe3C for the curious). If that steel is heated above 912°C (but not melted), something interesting starts to happen: the ferrite structure changes into austenite. Avoiding the deep weeds of metallurgy here, all we need to know is that austenite is a different cube of iron that is able to absorb carbon atoms, pulling them out of the cementite and trapping them inside the individual austentite structures.
Austenite is neat but unstable at room temperature without the addition of alloys (particularly nickel in stainless steel) that our blacksmith doesn’t have. If the austenite is allowed to cool slowly, it will slowly reorder itself, forming back into cementite and ferrite, often in a layering pattern called pearlite, which would, for the most part, put us back to where we started. But if the austenite is cooled very rapidly, it doesn’t have time to eject its absorbed carbon in an orderly fashion to form ferrite and cementite; instead the cubes of iron scrunch down into a ‘body-centered tetragonal‘ of iron which traps and essentially squeezes the carbon atoms inside. That formation is called martensite. This rapid cooling (quenching) essentially ‘freezes’ the steel in this state, whereas slow cooling would allow the martensite to transition back to austensite and from there back to ferrite and cementite.
Martensite is very hard but also very brittle. Steel with lots of martensite is thus also going to be very hard but also brittle and that’s a problem for most (but not all) applications. but if the steel is heated up again – not nearly so hot – this will cause some of the martensite, which is stable at room temperature but not above 200°C or so (that figure is approximate, I can’t find an exact figure and given how tempering works on a bit of a spectrum, that may be because it is really a range; this is why you have to very rapidly cool through the space between 900°C and 200°C, because your martensite is going to want to dissolve inside of that range) the martensite dissolves back to ferrite, cementite with a bit of austenite stuck in here and there. Higher temperatures cause more of this to happen, resulting in relatively less martensite and thus a relatively softer, more malleable steel. Because the temperature to which the steel is heated determines the degree to which the martensite is dissolved, the blacksmith can exert fine control over the characteristics of the final steel (and also, I should note, because high carbon steel is just generally stronger, tougher, harder and springier than iron, the blacksmith can achieve superior results to iron or low carbon steel in these characteristics).
Actually Heat Treatment
So how does the blacksmith utilize these steps to his advantage? Well the first thing to do is actually to normalize the steel. The reason is that all of this heating and cooling and metallic structure changes induce a lot of stress on our metal. In particular, martensite is both brittle (so it will break instead of bend) and less dense than austenite, so suddenly cooling a bunch of austenite into martensite is going to put a great deal of straing on the metallic structure of our steel. If we do that to steel that already has lots of strains from hammering (that is, work hardening that occurred as the iron cooled while being forged) it may well crack on us, which we very much do not want. Since all of the heat treatment processes are done and undone by heat, we have to do these steps after forging; we cannot heat the metal again or we will ruin the temper. Consequently, our steel must already be in its final shape, with as minimal stress as possible. So we begin by normalizing our steel to remove any existing strain and hardness.
Next is the actual hardening. The steel is heated up to around 900°C (the blacksmith will, as always, be gauging the heat of the iron by the color) to get that ferrite+cementite to austenite phase transition. Then the steel is quenched, rapidly cooling it, creating our martensite and getting a steel of maximum hardness.
Of course, rapidly cooling puts all sorts of strain on the steel, so the goal here is to cool as rapidly as possible without cracking the metal to achieve full hardness. Steels with higher carbon content will have more austenite (compared to ferrite; because there is more carbon to make the former) and should be cooled more slowly. Likewise, thinner sheets of steel can be cooled more slowly, but thicker objects need more rapid cooling (because they have a higher internal volume-to-external-surface-area, meaning that the interior can remain hot even as the exterior has cooled). The main tool the blacksmith has to control the cooling rate is the quenching medium: water cools quite fast, but oil quenches more slowly. This was known to the Romans as Pliny is aware of it (Plin. NH 34.145-6). A blacksmith is likely to rely on his own experience and the practices he picked up during his apprenticeship to know how to quench different kinds of objects and different steels.
Hardened and quenched steel is going to be far too hard and brittle for most uses (although some objects, like files, might be hardened and left in that state, since they do not need to sustain shock, but merely needs to resist abrasion). So now the steel is tempered, by heating it up again between 200°C and 330°C (all of these heats are using the same old forge; by controlling air input through the bellows and the exposure time of the iron, the blacksmith can finely control the heat of the metal, which he can gauge by its color) which will dissolve some of the martensite, removing the brittleness. After tempering, the tool is typically quenched again to once again ‘freeze’ the metallic structure in place at the desired hardness.
In a fascinating twist, the composition of the steel impacts the color of the oxide film that forms through oxidation on its surface through this process meaning that it is possible to gauge the temper of the iron from its color (though finished tools will have had this film polished off). Healy (op. cit.) gives the following chart (also reproduced in Sim and Kaminski (op. cit.):
|290-330°C||Blue||Saws, stone chisels, cold chisels|
|270-290°C||Purple||Swords, knives, woodworking chisels|
|250-270°C||Brown||Axes, wood chisels, shears|
|220-250°C||Yellow||Razors, turning tools, scrapers, engraving tools|
The lower the temperature that the steel is tempered at, the relatively more martensite survives inside of its structure, resulting in more hardness but also more brittleness.
At this point our steel tool has had quite a journey. It began its life as some ore and trees. The trees were reduced to charcoal and the ore was mined and then smelted, hammered into a billet and then into a bar. That bar was then carburized (if it was to be carburized and hadn’t been during the smelting process) and then forged into the shape of our intended tool. Finally, it has undergone a hardening treatment, either a limtied amoutn of cold working to induce work hardening if the carbon content is too low for heat treatment, or else it has been normalized, hardened, quenched, tempered and quenched one last time.
So we’re done right? Well, not quite. One thing almost always neglected in depictions of iron-working in fiction is finishing. Because we we’ve ended the process with is not a pristine, ready-to-use sword or tool. All of the heat has caused oxidation which has left a thin film of (brightly colored!) rust. Rust is very bad on iron objects – unlike copper which rusts protectively, iron rust encourages further rust. So that has to be removed, typically in the polishing process.
Moreover, our object is likely to have artifacts of the forging – the remains of the part of the bar the smith used as a handle, for instance – which have likely been forge cut off, leaving a knob that needs to get removed. Likewise, while any edges which need to be sharp have been drawn thin, they now need to actually be through to an edge or a point sharp enough to actually cut things. There may be other small adjustments that need to be made by removing metal as well. All of this is going to get done through filing and often quite a lot of it. Grinding stones might also be used, but a lot of this was done by hand (my sense is that rotary grinding stones are more common in depictions of the past than in the actual past, though they were certainly an important tool for any blacksmith). We almost never see this sort of finishing work in fictional smithies (or, for that matter, in period artwork, which much prefers to show the forging process), but it would have been an important and labor intensive task (likely done by apprentices rather than the master blacksmith).
And of course people like their expensive iron objects to look good. A forge tool the blacksmith makes for himself might be left in fairly rough, practical condition, but something that is going to be sold, especially prestige objects, are going to be carefully polished. In the case of pattern welded objects, this is also the stage where they would be chemically etched to bring out those wonderful streaky patterns so that everyone knows you spent a lot of money on your sword.
The object also needs to be protected from further oxidation (read: rust), which can be done in a number of ways. Rust is a huge problem for any carbon steel (or iron itself). While modern steel alloys are often quite rust resistant (although making armor or weapons out of something like stainless steel is, by the by, a bad idea, as it isn’t strong enough), high carbon steel rusts with depressing speed, even at room temperature in normal humidity, if not protected. Because rust forms at the surface layer of the iron or steel, rust prevention methods available in the pre-modern period mostly involved coating the iron or steel in something that would keep the iron out of direct contact with the moisture and oxygen in the air around it.
The most permanent solution might be to apply a metallic coating as either a liquid or as leaf using soft metals with low melting temperatures, like tin, silver or gold. The problem, of course, is that for iron intended to be at the business end of something – a sword, a hammer, a knife – those coatings are rather counter-productive and will wear on the edge with startling speed. Instead, armor might be blued or blacked. Here, the oxide layer formed during tempering is exploited (crucially, the oxidation during tempering forms Fe3O4, rather than normal rust’s FeO(OH); when quenched in oil, it absorbs the oil, creating a coating that will persist for years (assuming it is not polished off and is kept dry). Depending on how this is done (I am honestly not quite solid on the details myself) it produces a surface coating of an either dull-grey-blue color (bluing) or the classic gun-metal black-grey (blacking), which – so long as it isn’t worn, scratched or polished off – will resist rust.
Weapons and armor seem to have only rarely undergone this process. For blades and weapons sustaining impacts, this makes sense: regular use and sharpening would remove the finish quite quickly. But even armor, which might be blued or blacked, typically wasn’t, especially if it was going to be visible (that is, not under a textile like a coat-of-plates). Part of this is cosmetic – everyone likes bright, shining metal – but it also has a morale impact on the battlefield. For a soldier viewing a hostile army at a distance, one way he might be trying to gauge the quality of the fellows he is facing is by how much expensive, high-end metal equipment they have. A gleaming formation of shining steel and iron would thus be – and we are repeatedly told this by our sources – absolutely terrifying. So instead, things like armor, weapons and often tools were instead polished (to remove any rust) and then coated in oils (olive oil and fish oils both work quite well and seem to have been used historically) to stave off further rust. Such a coating would need to be regularly reapplied (…I can attest from experience…) which is part of why ‘shining armor’ was such a signifier of a good, diligent knight or soldier – you could see how carefully his equipment was maintained.
I don’t want to claim that we’ve covered quite everything here and so I want to start the conclusion by noting some important things we have not really talked about. I haven’t gotten into the role of markets, merchants and trade here because they vary so much by period. That said, the common image of swords being bought directly form the blacksmith is wrong in most periods. As earlier as we have evidence, in the Mediterranean, at least, we are hearing about professional arms dealers in places like Greece and Rome (and likely elsewhere as well). Tool sellers are harder to see, but must have been similarly common. Certainly we see lots of evidence that these items – being valuable and difficult to make – were used and sold and resold and traded and reused. A sword that began its life as a high-quality piece for a wealthy knight might end its existence as a cheap, second-hand weapon many decades later, after rust, damage and wear had their say. Not only were the finished products bought, sold and shipped, but so was the iron and steel itself. There was quite a lot of market activity in metals and metal tools (although compared to agriculture, such operations were a tiny slice of the overall pre-modern economy, perhaps never much more than 5% or so, to give a very limited sense of scale based on an estimate in Sim and Ridge, op. cit.).
Nevertheless, given this long process and all of the steps we have marched through to get to our finished tool or armor or weapon here at the end, I hope this explanation gives some sense of why iron or steel objects tended to be some of the most expensive things that a non-elite individual (and even some elite individuals) might own. Compared to similarly sized products in stone or wood, wrought iron and steel demanded tremendous investments in labor and fuel (which of course, demanded even more labor). Those workers, the miners, colliers, timber-cutters, furnace operators, barsmiths, blacksmiths, strikers, apprentices, all need to be fed and clothed and housed.
Leaning back on our series on agriculture, you can quickly imagine the impact that has on the structure of society. Almost all of our iron-workers (save perhaps our timber-cutters and colliers) are specialists who are not going to be providing those basic goods for themselves, which means that in order to employ them, the broader society needs to be producing surplus agricultural products (food, but also textiles) for them. That in turn brings us back to the idea of societies potentially having both high- and low-equilibrium points, because good iron tools can enhance the productivity of almost everyone: a good steel axe is much better at cutting down trees, a good iron plow is much better at plowing than stone, copper or wood equivalents.
Consequently, we can imagine two societies, on identical lands, with identical farming bases, but if one has developed an extensive iron-working industry (with improved tools leading to higher yields and lower labor requirements and thus greater efficiency) and the other has only a limited iron-working industry, the former is likely to be able to support more people at a higher standard of living than the latter. Moreover, just like with our plow-teams and fertilizer, the second society might find itself in a capital ‘trap’ where the absence of the tools and equipment needed to raise yields keeps surpluses low, which prevents the accumulation of the tools and equipment. I stress this because there is an assumption (occasionally even among scholars!) that because the basic biological yields of farming – how many seeds a wheat plant grows – were more or less static that then the productivity of the society must also be static. Certainly the efficiency gains from iron tools and plow-teams is nothing compared to the explosive productivity growth of industrialization, but nevertheless it could be significant on the smaller scale of pre-modern societies and have real implications for the living standards of the broader population.
And that is our series – for now – on iron production. I expect to add at least one addendum to this series, covering pre-modern Chinese cast iron production as well as Indian crucible steel (‘Wootz’ steel). That said, I am taking next week off from the blog (I have a mix of grading and writing that is time sensitive and needs to get done) – there will be a post next Friday, but it will be me ‘re-surfacing’ some of my older posts that I think are good but perhaps didn’t get quite so much reading the first time around.