This week, as an addendum to our four(and a half)-part (I, II, III, IVa, IVb) look at pre-modern iron and steel production, we’re going to look at two alternative regional processes, where they fit into the entire iron production process that we detailed already, how they worked and what kind of product they produced. In particular, we are going to look at the early production of cast iron in China, and the production of a form of crucible steel, called wootz, in India (and South Asia more broadly).
First, some necessary caveats. Neither of these methods of iron production were practiced in my core area of research (the greater Mediterranean world during the Hellenistic and Roman periods), although wootz steel was imported in small quantities during that time and a handful of cast iron goods produced in China also made it to the broader Roman or Hellenistic world (though very, very few). Consequently my knowledge here is much thinner and I am much more reliant on reporting the work of other scholars. Moreover, some of the scholarship on these topics is in languages I do not read, so I am also reliant on other scholars for that. As such, this is going to be a much more general discussion of these processes and I am going to be a bit more cautious about what I say about them. At the same time, I simply don’t have the information to talk much about the details of the lives of the iron-workers here themselves, like I could in the previous posts, so this will mostly focus on process and product, rather than people.
That said, I thought this was still worth doing because it seems like there is quite a bit of unhelpful ‘mystique’ surrounding both. On the one hand, a lot of the discussion of the production of cast iron in China fails to really put the innovation both in the correct spot in the iron production process but also in the correct context in terms of the development of iron-working technology. On the other hand, much of the popular discussion about wootz steel in India treats it as this quasi-legendary ‘lost’ and mysterious production process when it is, in fact, a fairly well understood method of producing high-quality steel.
And of course, as always, if you want to support my essay production process, you can do so via Patreon and by sharing to my work to a wider audience (but please do not copy-and-repost these essays; link to them instead). 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 (and I promise fewer political musings now that the electoral götterdämmerung has come and – heavens I hope by the time this post goes live – gone).
(Subsequent edit: Due to some election-related trolling coming from off-site, I had to prune back the comments to the original topic. Apologies for those whose comments were pruned (except the troll) but I wanted to keep us on topic and away from the abyss of endless political carping.)
Cast Iron in China
Now I’ve stated several times during this series that generally speaking the bloomeries used to smelt iron in the pre-modern period were not capable of reaching the high temperatures (1538 °C) required to actually melt iron and instead produced pure metallic iron through solid state reduction (that is, the bloomery process) along with, essentially, melting everything that wasn’t iron out of the bloom. And for most of the world in most of the pre-modern period, this is true.
China appears to be the odd exception. I should note that this is an area of scholarship that is still quite active – dates are still in flux (shifting back as new discoveries push back the earliest dates) and the exact nature of the transitions from bronze-working to iron-working in China and the degree to which the bloomery processes we’ve discussed served as an intermediate stage is still debated, as far as I know. A fair bit of the discussion of the question is, unsurprisingly, in Chinese, which I cannot read. So I am doing my best to provide a summary of what I know.
Cast iron in China seems to become widespread during the Warring States period (453-221), though the technology itself probably had its genesis in the earlier Spring and Autumn period (770-454) and it becomes truly ubiquitous during the early Qin-Han Period (221 BC – 220 AD). The general working hypothesis based on the evidence is that the cast iron product emerged not out of bloomery iron production (which seems to have been used in China side-by-side with cast iron production for the first couple of centuries) but instead out of the furnace techniques already in use for casting objects in copper-alloy (read: bronze), until the cast iron process became able to produce a high enough quality iron product that it supplanted bloomery production, probably in the last century BC. Again, those dates are very approximate and seem to me to still be significantly unsettled, going by the range I’ve seen in scholarship (the most recent I could find on this was “Everything Old is new Again? Rethinking the Transition to Cast Iron production in the Central Plains of China” in Journal of Anthropological Research (2014), so those are the dates I have gone with; Craddock and Tylecote both have somewhat later dates).
The basic interaction of producing cast iron centers around one oddity in carbon-iron alloys: as you add more carbon to iron, the melting point of the resulting mixture drops. At 4.3% carbon content, the melting point of iron drops to just 1150 °C, which as you will recall, is very much within reach for pre-modern furnaces worked by bellows (and China seems to have generally had better, hotter furnaces than contemporary pre-modern Europe to boot). Adding small amounts of either sulfur (typically from coal being used to supply the carbon) or phosphorous can lower the melting point further, although these can also make the metal brittle. At that point, large, bellows-fed furnaces can heat the iron-carbon mixture up to the point where it melts and becomes sufficiently fluid to cast (slag and other impurities, being lighter, can be drawn out with a flux and tapped off). The result is cast iron (also called ‘pig iron’).
The problem, as we’ve discussed earlier, is that cast iron (which, to be clear, iron produced this way is still cast iron, even if you aren’t pouring it into a model; it is a type of iron as will become plain) comes out of the process with unfavorable characteristics, particularly in strength, brittleness and malleability – in short, it is too hard, too weak, and too prone to break and crack rather than bend. The carbon content of cast iron is, of course, very high: above 4%, the carbon makes the iron much too hard and inflexible, causing it to become extremely brittle. In particular, if the silicon content (which is likely to come from bits of slag melting and dissolving into the iron at the high temperatures) is low and the iron is cooled fairly quickly, it forms into cementite, which is hard and brittle. Cementite isn’t bad in and of itself – we saw this with heat-treated steel, where just a bit of cementite among the ferrite offered improved strength and hardness. But this is not just a little cementite, it is lots of cementite. In excess of 60% cementite, what is sometimes called ‘white’ cast iron. At these levels, the amount of cementite leaves the cast iron very hard and very brittle, likely to break and actually difficult to cast effectively (because it is likely to crack from thermal stresses as it cools).
More slowly cooled cast iron with high silicon content, can come as ‘grey’ cast iron, which has less cementite, but has instead pockets of graphite – little pockets of cure crystalline carbon which couldn’t be absorbed into the iron – which have very little strength and essentially functions as cracks within the iron, reducing the overall toughness considerably. Craddock notes that early Chinese cast iron tends to be grey cast iron, but – curiously – with somewhat lower silicon contents than we’d expect for grey cast iron. So for almost any tool or weapon use, our cast iron is functionally worthless: too brittle to sustain meaningful stress. But we’re not done.
The key technique that Chinese ironworkers seemed to have developed here is malleablisation. By heating the iron to melting temperatures (or just below) in an oxidizing environment (which is to say, open to the air) the ironworker can cause a few things to happen. First, the oxidation process begins to draw off the carbon from the iron (again, you will remember the tendency of heat and air to decarburize iron from our discussion of making steel). At the same time, the graphite which originally was in long, thin flakes begins to form into more compact round shapes which cause less strain on the metal (there is also just less of that graphite, because it is made of the carbon that the air is drawing off in the decarburization process). If the iron is kept hot and open to the air long enough, it can either be decarburized to steel or even all of the way to wrought iron, making it usable for tools, weapons, armor and the like.
(I should note that somewhat different processes for removing carbon to make cast iron workable as wrought iron, called fining, emerged in late medieval Europe. This was done at a finery forge (we start to see these as early as the 13th century AD in Europe and involving breaking cast iron into smaller chunks and then heating these in an oxidizing environment so that the iron would melt and excess carbon (in the form of graphite) would burn out, leaving a bloom below where the iron had been. That technology had matured enough to produce armor and weapon’s grade steel by the 15th century, which is, to be clear, quite a bit later than the first century BC when the method was mature in China.)
And so by the start of the Qin-Han period (221BC-220 AD), we see a wide variety of objects (tools, weapons, etc.) that are made out of cast iron that has been, to some degree, malleablised. The amount of malleablisation that has taken place in these objects tends to vary wildly though, with microstructure (and thus metal quality) differing quite a bit piece to piece, suggesting, as Craddock puts it that “the complex and time and energy consuming treatments [read:malleablisation] were often not carried out or failed to have the desired result” (Craddock op. cit. 269). Which, given that the process of malleablisation often required keeping the iron hot in the open air for hours if not days, given the fuel demands, it is not surprising that sometimes lower-quality metal was produced by cutting the process short. By the first century BC, Chinese methods of malleablisation clearly improve, with solid state decarburization becoming reliable enough in producing steel and iron from cast iron that bloomery-smelting was abandoned in favor it.
Now it is important to note which parts of the production process that cast iron production replaces and which it doesn’t. For very low grade iron products – things that can actually be directly cast – cast iron production might actually replace both the smelting (that is the bloomery) and some of the forging work of producing an iron object. But for most iron objects – including nearly all tool and weapons – cast iron production essentially replaces only the smelting process, while malleablisation would replace the carburization process with a decarburization process. If the iron is kept liquid during the malleablisation process, it might be possible to cast it into the general shape of the intended object, but you would usually need a blacksmith to do some finishing work. I am not clear on how often malleablised iron would have been cast into its near final shape as compared to being produced in bars to be forged.
Placing Chinese Cast Iron In Context
All of which is striking because the spread of the Chinese method of cast iron production seems to have been fairly limited. The technology clearly seems to have spread into South Asia, where we know that cast iron was sometimes mixed with bloomery iron in the crucible process (below) to produce steel – South Asian ironworkers were thus using cast iron to create crucible feed-stock, but not as a final product. But while the technology was available, South Asian iron-workers stuck with bloomery processes for the production of wrought iron, rather than using the cast iron and malleablisation process developed in China, despite the fact that they clearly knew how to do it. Likewise, while Chinese iron working clearly had a lot of influence on iron production in pre-modern Korea and Japan, they too continued to rely on solid-state (read: bloomery) reduction methods for producing iron and did not adopt the cast iron process.
Why? Well, the key thing to note here seems to be that the malleablisation or fining process was long and fuel intensive. As opposed to the bloomery processes, I don’t know as much experimentation with this, but I have seen fining times quoted in the range of several days to properly turn cast iron into steel or wrought iron – time during which the iron must be kept open to the air (so it is cooling on its own) and also at least nearly molten. Indeed, the one method, to keep the carbon content relatively uniform, was to keep the iron completely molten and to stir it regularly as it decarburized (which might also have shortened the time required). The fuel demands of that method would have been enormous.
Moreover, casting iron and then malleablising it could produce an inferior product – steel produced this way tends to have higher sulfur, phosphorous and silicon contents in it, which are generally elements you want to avoid in an iron-carbon alloy if you can; cast iron Chinese artifacts seem to generally have lower percentages of these elements in them than, say, comparison to early European cast iron would suggest, but lower is not zero. The reason is, unlike in the bloomery process, which uses nearly pure-carbon charcoal and reduces away all of the silicon and many other trace materials, in an a cast iron process, mineral coal is often used which may have trace amounts of sulfur and other trace elements in the ore, instead of being reduced away, can be absorbed by the melted iron. The crucible process (below) avoids this problem by reducing the iron ore first (through a bloomery process) and then melting it second, once it is relatively pure. This may explain why, despite the potential for the malleablisation process – if performed with the iron fully molten – to produce iron with functionally no slag and very uniform carbon content, Chinese malleablised iron never gained the reputation as a superior product that wootz so clearly enjoyed.
As a result, one thing that – to my knowledge – we do not see much of are Chinese iron-working products traded long distance, in stark contrast to South Asian wootz steel (which was, among other things, a valuable trade item, in China, suggesting that it was known to be superior to the local product). I know of only a couple such objects, for instance, in the broader Roman world – all cast iron cauldrons (discussed in Tylecote, op. cit., 325-6), which is striking given that the Romans were importing quite a lot of other things from China (particularly silk) and also were quite avidly importing wootz from India (so it wasn’t as if they weren’t in the market for foreign, good quality steel). There was, of course, much more diffusion of Chinese iron products around China (but most of the examples I see are of Chinese iron working diffusing into the Steppe, where there is essentially no competition, since nomadic peoples don’t generally do much metal production), but Chinese iron goods don’t displace local products in places like Japan, Korea and South-East Asia, suggesting that transport costs outweighed any production advantage offered by the different process.
Consequently, as far as I can tell, the Chinese cast iron process – unlike the much later Bessemer Process – wasn’t radically cheaper, faster or more efficient than bloomery processes (although it does seem to have been somewhat better) and it may have, on the average, produced a slightly inferior end product in many cases. I have to stress that this is a very rough sense of the matter, because I haven’t seen any effort to gauge with any kind of certainty (in contrast to experiments with both bloomeries and crucible steel) the productivity and fuel demands of Chinese cast iron processes, so this is mostly reasoning from the lack of diffusion.
All of which matters for where we understand the Chinese cast iron production system to fit in the development of iron working. Often, Chinese cast iron production is discussed quite breathlessly, with implications that China was massively ahead of the rest of the world in metallurgy. It is important here to be careful with some of the statistics used. To single out just one example, J.L. Abu-Lugbod in Before European Hegemony (1989) repeats Robert Hartwell’s estimation that “the tonnage of coal burned annual in the eleventh century for iron production alone in nothern China was “roughly equivalent to 70 percent of the total amount of coal annually used by all metal workers in Great Britain at the beginning of the eighteenth century” (Hartwell, 1967: 122)” offered as a sign that China was producing “at a scale not equaled anywhere in the world until the Industrial Revolution” (324). It all sounds very exciting.
But there are all sorts of problems with this assessment. First, trying to assess iron production through coal usage is very unwise, since the amount of coal (or charcoal) that might go into a given amount of iron will vary wildly depending on the efficiency of the method used. It can be little doubted, for instance, that the Bessemer process, which completes a decarburization heat of iron typically in less than 30 minutes, uses far less coal per unit of steel created than Chinese malleablisation which, as noted above, had to keep the iron heated – often to melting – in open air conditions, for days (of course, the Bessemer process wasn’t developed until 1856 and so cannot figure into Hartwell’s calculation, this is meant to illustrate the point; going by iron prices, the crucible processes in use in Britain in 1800 were about 6 times more expensive than the Bessemer process, which is a big difference, but not so big as the difference between a few days spent decarburizing or thirty minutes).
It is also, by the by, pretty absurd to directly compare coal utilization of an empire of perhaps more than a hundred million people (in c. 1100) to an island of 10.5 million. I would be extraordinarily surprised if any large pre-industrial, agrarian population was out-producing another by a factor of ten – prior to the industrial revolution (which doesn’t really hit steel production until after 1800), the efficiency gains for that kind of thing simply aren’t there. For comparison, the typical back of the envelope estimate for the increased production of the Roman Empire at its height as compared to the period before or after is typically in the neighborhood of a 25-50% increase; it is surely not anything like the 1,000% necessary to get 10 million people to out-produce 100 million. It is certainly fair to suggest – and I think the evidence does suggest this – that Chinese iron production in the 13th century was more efficient than British iron production in the 13th century; to proceed beyond this to grand conclusions seems profoundly unwise given how limited the evidence is.
Now that hardly means that Chinese cast iron production was not an achievement. This was a technical feat that contemporaries in Europe would not achieve until the 15th century. Cast iron production does have its advantages and it is absolutely fair to point to cast iron production as an example where China had a technological edge. When European firearms like the breach-loading swivel gun arrived in China, the locally made Chinese versions were often superior to the Portuguese originals they were copying because they could be cast in a single piece, rather than built up using hoop-and-stave construction. But it seems better to understand Chinese cast iron production, and iron production more broadly, as one within a family of pre-modern iron production methods, with advantages and disadvantages (perhaps more of the former than the latter, but certainly both), rather than as an early version of modern steelmaking. it is striking, given how rapidly both pre-modern ironworking and modern steelmaking spread that many of China’s neighbors look at the – again, probably somewhat more efficient – cast iron process and said “nope, not for me,” and kept saying that for centuries (even as silk, Buddhism, tea, gunpowder, and all sorts of other things quite cheerfully crossed those same cultural and geographic boundaries).
This becomes, I think, clearer when we compare this to the contemporary method of producing steel in India, which – while perhaps not as efficient as Chinese cast iron – did produce a pretty clearly superior product.
Crucible Steel in India
While Chinese cast iron products do not seem to have been distributed widely over trade networks, beginning at least as early as the Roman period, Indian steel (which the Romans misunderstood as being from China, since it arrived on the same trade roots as their silk) was. This steel, generally now called ‘wootz‘ in the literature (but see below on naming) is a kind of crucible steel.
The method of producing crucible steel (which is developed in Europe only much later, in the 18th century) begins by using fairly high quality wrought iron as feed-stock (that is, iron produced in the bloomery process), so we should keep in mind that the crucible process replaces the cementation or case-hardening process in turning iron into steel, not the smelting process (as compared to cast iron above). Indeed, it seems that in many cases, wootz steel was produced from a mixture of wrought and cast iron, the latter either imported from China or later produced locally. That iron was broken up into smaller chunks and then mixed with organic materials to provide additional carbon. Plant materials (leaves, wood, fruit skins) were the most common; interesting charcoal seems to have been used only rarely. The knowledge of the ratios and mixtures in this process must have been learned through a long period of experimentation, especially since the actual process inside the crucible could never have been observed.
That batch was then sealed up in crucible. The trouble here seems to have been finding a refractory (read: heat resistant) substance which could withstand the intense temperatures used in the process without either melting or cracking from the thermal expansion. What seems to have been used was a clay vessel, mixed with rice husks – during heating, the husks trapped in the thick walls of the crucible would char away, creating little voids that the hot clay vessel could expand into, reducing thermal stresses. Of the whole process, I must say, this use of two common materials to make an uncommon one struck me as one of the most ingenious things I had ever heard of. The crucible was sealed completely to keep out all oxygen – often the lid was sealed with a different sort of clay that melted at lower temperatures so that, when heat was applied, its partial melting would create a perfect, air-tight seal as it melted over the edges of the crucible.
The sealed crucible was then fired in a charcoal fire, worked by bellows which would push the temperature probably in excess of 1400°C (which is to say, a crucible furnace actually burned hotter than a cast iron producing furnace). As the iron absorbed the carbon in the carbon-rich materials trapped with it (as they would rapidly be reduced to charcoal in the heat), it would lower the melting temperature – but not nearly so much as with cast iron, because not so much carbon was absorbed. Consequently, the furnaces heating the crucibles would need to hit higher temperatures than in any of the processes we have looked at so far. Once the iron melted, all of the slag – everything that wasn’t the iron – would float to the surface of the melt (because slag is a lot less dense than iron) and form a crust on the surface, removing nearly all slag inclusions from the resulting mass of solid iron. By varying the amount of organic material, the iron-worker could carefully control the carbon content.
Firing times seem to have generally been long, though there is a lot of variance in the evidence. Some report firing of a single hour, but in most cases, the crucible had to be kept at these high temperatures for many hours, up to a full day. Once this was completed, the crucible was broken open and the steel left to cool – typically very slowly, as the slow cooling would allow the wootz to develop the proper crystalline structure. This slow cooling process was crucial to produce the very high grade steel that was meant with the term wootz and Craddock (op. cit.) notes that Persian merchants trading in the steel would carefully monitor this part of the process to ensure they were buying that superior product for west-bound trade (to the Near East and beyond). The slag, which would solidify at the top of the mass of iron, could be broken off with hammers.
Wootz produced this way would typically come out of the process with 1-2% carbon and very low slag-counts, making it about perfect for sword-blades and the like. So, while Chinese casting processes didn’t generally produce a superior steel, the wootz crucible process really does seem to have done so. It is clear that the crucible process was used to produce other grades of steel as well though: rapid cooling would produce a much harder but less flexible steel (generally an inferior, but not useless product), while products that required low-carbon steel could be intentionally decarburized during forging and annealing. A skilled ironworker could also probably vary carbon content intentionally by tinkering with the ratio of wrought iron to sources of additional carbon in the crucible, be that cast iron or organic materials.
It is not entirely clear when Indian iron-workers began producing true wootz. There are a handful of sources tentatively noting high quality iron coming out of the East as early as the third century BC, but the evidence is thin and disputed (on this note Craddock, op. cit. 278). Instead, the first good evidence for wootz in a literary source seems to be Pliny the Elder’s Natural History (34.145, written in the 70s AD) where he mistakes the point of origin (calling the iron ‘seric’ (= Chinese) rather than Indian). By the third century CE, we have Greek sources (Zosimos) actually describing – with surprising accuracy – the process for the production of wootz and plenty of evidence for this high quality steel moving into the Near East and the Roman world through trade.
This gets us into some of the ‘mystique’ around wootz which I want to work to dispel. Wootz is – along with early medieval pattern-welded steel – often held up as some sort of super-metal, often with the assertion that the production method is lost (clearly not, reading above!) or that it produced a higher quality steel than modern steels. And certainly, at one time, wootz was mysterious to Europeans. As noted above, the Romans didn’t quite know where wootz came from and assumed that – since it arrived on the same trade routes as their precious silk – it must also come from China. Medieval European writers, seeing that wootz arrived to their world from the East sometimes called it ‘Damascus steel’ because, as the steel traveled west on the trade routes, that was where they would have first become aware of it.
All of which have contributed the air of mystery around wootz. But being mysterious to Europeans in the Middle Ages is not the same thing as being mysterious generally or still being a mystery now. Wootz production was not mysterious in South Asia, where there were multiple attested centers of production, including in Hyderabad, Sri Lanka, and Mysore (different areas were production centers at different times). Nor was wootz terribly mysterious in the Islamic world. Middle Eastern traders knew exactly where it came from and by the 12th century, Islamic writers are recording good descriptions of the process, which was beginning to be used in the Middle East by that point (Craddock, op. cit. 279). Nor did wootz production mysteriously vanish either – there is evidence for extensive production of crucible steel in Hyderabad in the 17th century; early 19th century accounts by British and Russian writers also describe the process It was never lost! D. Mushet patented what was, as Craddock notes, “clearly a copy of the Indian process” in 1800 (Craddock, op. cit. 283) in Britain. My impression, though none of my sources quite say this directly, is that the remaining Indian steelmaking industry subsequently withered away due to competition with British steel and (perhaps more decisively) a British desire to keep steel production (with its obvious implications for weapon production) in Britain and not in the (potentially rebellious) colonies.
In terms of the quality of the steel produced by this process, wootz was clearly a high quality steel, with relatively uniform carbon content and extremely low amounts of slag inclusions. A steel blade made of wootz would out-perform locally manufactured blades in Europe or the Middle East. But that doesn’t make wootz some sort of lost super-metal either. The modern high-carbon mono-steel used in high quality replica weapons (‘high quality’ being key here, since many modern replicas are made very poorly) is also generally uniform, low-slag content steel, not dissimilar from the qualities wootz would have had. And a good pattern-welded blade produced in the Middle East or Europe during the Middle Ages would not have been very much inferior to a similar sword made of wootz – although wootz does seem to have been noticably better, if you could get it.
So while wootz was valuable enough to be worth trading over long distances and was probably the best option for a high quality sword, it was neither mysterious, nor some sort of lost super-metal, just a (very) high quality form of steel. And while it was probably often the best option for a very high quality steel weapon, it was not so much better that local steel couldn’t also do the job, such that in many cases having a sword made of wootz had more to do with the prestige of it (the wavy pattern being a clear indicator that the expensive material was used) than its superior characteristics.
And that is that for our series on iron. I know there have been some requests for a run-down of modern methods of steel production (some of the early versions of which I alluded to above), but I think that I am probably not the right person to write those, as my expertise is in pre-modern production. Moreover, there is plenty written and readily accessible about modern steelmaking and I am not sure what I would usefully add.
Next week, something different!