The Ways of the Ancients

It appears that an ancient smith in Persia was making steel that was close, in today’s terms, to what we’d call “stainless steel.” The question is whether that was lucky, or intentional, or both.


Chromium steel, commonly referred to as stainless steel, is thought to be a recent manufacturing innovation, but new evidence suggests ancient Persians stumbled upon an early version of this alloy some 1,000 years ago, in what is a surprise to archaeologists.

Ancient Persians were forging alloys made from chromium steel as early as the 11th century CE, according to new research published today in the Journal of Archaeological Science. This steel was likely used to produce swords, daggers, armor, and other items, but these metals also contained phosphorus, which made them fragile.

“This particular crucible steel made in Chahak contains around 1% to 2% chromium and 2% phosphorus,” Rahil Alipour, the lead author of the new study and an archaeologist at University College London, said in an email.

Back in the “good old days” smiths engaged in back-breaking labor to experientially figure out the properties of various metals. It’s not like nowadays, when you can go to New Jersey Steel Baron and order up a stack of 1095 high carbon steel 1/4″ x 2″ x 48″ and it arrives in a FEDEX truck a couple days later. Ancient smiths were melting down material they collected in various places, adding their own mix of special ingredients, and seeing how it performed. I was interested to see that the Persian steel had a lot of phosphorus in it, too – they got the chromium right but should have replaced the phosphorus with vanadium. But, at that time, there was no “vanadium” that anyone knew about. For sure, if an ancient smith had started making chrome/vanadium steel, their work would have been considered magical – except it wouldn’t forge-weld properly to anything.

Nowadays, smiths like myself refuse to work on “mystery metal” because you can spend hours producing something that looks great but snaps in half, or doesn’t hold an edge – and you have no idea why. I’ve talked to smiths who make knives out of old saw blades but even that is a crap-shoot; some saw blades are 15N20, others 1095, and more are “whatever went in the crucible that day” but they work as saw blades because they have plasma-deposited carbide on the teeth which means that the rest of the blade doesn’t have to do anything beyond “hold the whole thing together.” That, in a nutshell, is what pattern-welded blades, like I make, do: I can make a blade with welded cable on the faces, surrounding a chunk of 1095, and my whole problem boils down to making sure the 1095 ends up where the edge is going to be. My suminigashi brew is a bunch of stuff that no smith in their right mind would use for a blade, but it’s great for the facings of a blade because it’s tough stuff and can help hold the blade together.

By the way, I was talking with another smith about how hard it is to drill a hole through my suminigashi and he pointed out that what’s probably going on is that the transitions between the carbon and mild steels are forming carbide crystals in there. And, as he pointed out, “drilling carbide with a carbide drill means you use a lot of drills.” This is stuff that, today, we sort of kinda mostly understand, but a smith in ancient Persia would have just noticed that this particular batch of steel was particularly shiny and resisted rust well, so he tried to make more of it.

some Iranian wootz, 13th c

Then, as a thread through the entire story of Persian steels, is wootz steel. Modern smiths get into fist-fights about wootz. See, wootz is the original “damascus steel” (Damascus being a city in Persia, right?) and there are wootz blades of Persian origin going back to around 800AD when Persian smiths started getting their hands on steel from Northern India. It appears that they probably threw it in a crucible together with other stuff and, when they broke the crucible open and started refining the metal into a bar, they noticed a weird “watering” pattern. Basically, what was happening is that the different impurities in the steel (vanadium and manganese, mostly) made it separate slightly as it cooled; you got these treelike shapes (also referred to as “dendritic steel”) in the steel which were different ductility and hardness from the rest. Today, we get a similar effect by layering different kinds of steel together, and quenching them differentially, but the original wootz had its watery grain just by the accident of how it was made. I imagine that the smiths who first saw that were very excited, and spent the rest of their careers trying to refine and reproduce that kind of mix. Today, there is a small hard corps of smiths who spend their weekends dancing and drinking around propane/oxygen fueled vertical smelters full of a bit of this and a bit of that; i.e.: Joe’s secret wootz mixture of 3 parts disc brake, 1 part 1095, 1 part semitrailer axle, and some powdered vanadium thrown in. To me, it has the feel of a holy ritual that calls out across the years and miles.

------ divider ------

Making steel was a critical military technology until the gunpowder age (and, even then…) and it’s fascinating that, as a great military technology, various pieces of steel found their way all over the place. There are 13th century viking swords, unquestionably forged by vikings, but the metal appears to be of Chinese or Persian composition. Scientists can track this stuff. It’s so damn cool.

By the way, pre-1940s steel is valuable in its own right. Sunken WWII battleships are valuable (I’d love a chunk of the Tirpitz, seriously) in their own right. Hmmm… Let’s do it as a quiz. Make your guess as to why, and the answer is ROT-13 below:

nsgre gur 1940f nyy fgrryf fzrygrq pbagnva vfbgbcrf sebz ahpyrne nveohefgf. vs lbh ner znxvat n fhcre-frafvgvir zrnfhevat qrivpr, lbh jnag gb hfr fgrry sebz na byq fuvc gung qbrf abg pbagnva vzchevgvrf.


  1. aquietvoice says

    Hmmmm, the only thing that was present in pre-1940’s material that you can’t get now is probably…. no contamination by nuclear fallout, so that’s my guess, but I can’t figure out how that would make it valuable. Not a lot of contamination would be included in steel in any case, and none of the ingredients are really vulnerable to contamination.

    More specifically, I know that carbon-14 was at its lowest % ever in 1944, when it was diluted by fossil fuels but not raised by nukes…… but if you wanted low carbon-14 steel you could just source your carbon from something old – like fossil fuels, so that can’t be it.

    Perhaps something magnetic then? But that’s too easy to control externally – and I can’t think of any effect you can get by putting something in earth’s magnetic field for a long time that you couldn’t get with a strong field in a short time.

    I mean, I can’t imagine anything else that would alter what “iron with inclusions” is, but there’s always something extra to learn about the world.

    Heh, like the time that people figured out how to alter the rate of nuclear reactions (ok, a very specific nuclear reaction) with chemical reactions.

    (Get beryllium, choose an isotope that loves electron capture, muck around with the electrons to increase or decrease the overlap with the nucleus – ie. 2s-2p transition. Bam! Chemically controlled nuclear reactions….)

  2. astringer says

    * * resonance?

    I really enjoy these smithy entries. An old relaxation hobby has been to think how (if at all) useful I could be if thrown back in time to various ages, with no practical skills but some common knowledge (nothing fancy). These stderr entries show just how useless we would all be without the years of trial and error experience despite knowing, say, atomic theory.

  3. says

    Making your own wootz or tamahagane is just fun, nothing more. Even forging damascus is just for the fun and looks. No patterned damascus steel can hold a candle to anything made from mono steel properly chosen for the task at hand, or perhaps a san-mai/ni-mai from mono steels.

    Ultimately it is fun to make your own stuff, and discover and re-discover things for yourself.

    All blade makers that I know of, including myself, started making blades because they liked to do so. If we wanted to make boatloads of money, there are far more reliable, but decidedly less fun (for me, anyway) ways, to do so.

  4. komarov says

    There are three pretty generic reasons I can think of why old material is better than the new stuff.
    1) Restricted (or depleted) ingredients, e.g. some toxic additive that is no longer permitted or acceptable. That seems to happen occasionally, e.g. with solder, where modern lead-free solder apparently isn’t quite as good as ye olde (more) toxic stuff. Or asbestos: fantastic stuff that simply had to go everywhere until it suddenly didn’t anymore.

    2) Modern manufacturing methods can’t deliver the same quality but dominate because of economics, so the good but hard to make steel is no longer being made because no one would pay for it. (2b: All the people who used to know how to make it are dead)

    3) Capitalists figured they could make more money with less quality. That’s sort of 2c, but not quite, because there’s a difference between greed and economics. One is being circumspect, the other being in it for one’s own gain at the expense of everyone else.

  5. kestrel says

    The Partner sez: There was not as much background radiation pre 1940, so the old steel underwater has not been exposed the background radiation we have nowadays. That matters if you are trying to make an instrument that measures things like radioactivity. In southeast Asia the sunken old ships have been all salvaged (stolen in some people’s minds) for such purposes.

    (I hope I have accurately got this lecture I just received down to just a few sentences.)

  6. Reginald Selkirk says

    Once upon a time, about three decades ago, it first came to public consciousness that lead from lost fishing tackle (e.g. sinkers on snagged lines) was polluting the lakes in Minnesota, working its way into the food chain, etc. So environmentally aware persons approached the fishing tackle industry and asked, “Can you help us out here?” The fishing tackle industry was not the most scientifically & technically aware crew. “Gosh,” they replied. “That never occurred to us. I guess we could use something other than lead. How about cadmium?”

  7. Ketil Tveiten says

    Most of the Tirpitz is used as covering plates for road work here in Norway and probably elsewhere as well. Some guy bought the salvage rights after the war for cheap in return for getting rid of the damned wreck, and sold it as scrap

  8. Pierce R. Butler says

    Making steel was a critical military technology until the gunpowder age (and, even then…) …

    Will our esteemed host eventually make the transition to gunsmithing and cannon barrels in keeping with tradition?

  9. says

    I expect that the reason 1940 and earlier steel is prized is for some of the same reasons that nuclear and particle physicists, when building ultra low background setups want lead from Roman shipwrecks. Because the nuclear weapons detonations, be it in anger or for testing, from 1945 onward contaminates everything with various isotopes like 90Sr, 137Cs, etc.
    Additionally, pre-1945 steel will not suffer from 60Co contamination either. That is not from the weapons; because the half-life is only ~5.5 years so any created in the tests has decayed significantly and because any 60Co from the weapons was formed by activation of stable 59Co by neutrons instead of being one of the actual products of fission which means it is produced in lesser quantities than the fission products themselves. 60Co in steel comes from medical and industrial sources being unknowingly (or deliberately and irresponsibly) recycled into it. Since use of radiation for industry and medicine was mostly just X-ray tubes prior to 1945 (hell, probably prior to 1960); steel from before that time should be relatively free of that contamination too.
    Nuclear physicists and particle physicists also like roman shipwreck lead for two other reasons.
    1: Because the naturally occurring relatively shortlived radioactive isotopes from the “bottom ends” of the 235U, 238U, and 232Th decay chains that are normally “pulled” into the lead during the smelting process have had more than enough time to die away.
    2: Because it is at the bottom of an ocean or sea it will have had a fair amount of water above it shielding it from cosmic rays. This helps reduce the number of times high energy particles from the cosmic rays from creating additional naturally occurring radioactivity in the lead (be it from high energy muons or neutrons disrupting nuclei in a collision or from a neutron that has managed to thermalize, slow down, bumping into a nucleus and being absorbed, making it radioactive.)
    I diliently wrapped the atomic masses of each of the isotopes in <sup>A</sup> tags… it did nothing. sigh

  10. says

    @aquietvoice #1
    I very deliberately didn’t read any comments prior to posting mine. But you are right it is due to the lack of fallout being incorporated into the steel.
    Though the 14C is a bit of a red herring, that isotope is the second lowest energy beta emitter on the entire chart of the isotopes (to see this go here and click the little “button” close to the top that reads “Q_beta-” that will color the chart by the amount of energy that would be released if the nucleus underwent Beta decay a.k.a. the Q-value of beta decay.) It’s the long lived fission daughters that get pulled into the smelting of just about any metal. These contaminants are small, they don’t really affect the metallurgical quality of the metal (as far as I know), but they do make building certain scientific instruments etc with that metal more challenging because it has higher backgrounds.

  11. komarov says

    Re: Reginald Selkirk (#6):

    Uranium is also pretty dense. I foresee absolutely no problems there.

    Re: Pierce R. Butler

    “Will our esteemed host eventually make the transition to gunsmithing and cannon barrels in keeping with tradition?”

    Ultima Ratio Ranum

  12. lorn says

    IMHO it is sometimes difficult for us to understand exactly how arduous it was for ancient craftsmen to really understand anything and to be able to confirm and make usable what they were learning.

    First, they had very little theoretical understanding of what was going on. They didn’t have the concept of chemical elements. Likely they used a mental matrix of elementals (earth, air, fire, water) and some ideas of affinities, or antagonisms, and some concept of material experience and stories. They didn’t know anything about carbon of vanadium as such.

    Of course even those hard won understandings were likely set against a background of Gods, superstition, politics and for-profit tradition. If a batch of iron had interesting and useful properties they would, without any concept of experimental science, be hard pressed to know if the critical difference was the rocks used or, perhaps, it differs because you wore a green tunic instead of your blue one, or perhaps it happened that way because your daughter stopped menstruating, or, perhaps it was that beer you poured over an idol’s head in a drunken revelry. When you don’t know about the relevant categories of a physical world, and you still have one foot in metaphysics, really knowing things is tough.

    Point being it was mighty hard to stay focused and think clearly about just your work as a blade smith.

  13. says

    Today’s Syria was part of Persia in the 11th century, which is the time I am talking about – when Damascus steel was a big deal.

    Geolocating ancient tech is weird – the Syrians would not have been trading and incorporating steels from India on their own, that was a result of being part of Persia. I don’t think we should say that the Roman coliseum in Arles is actually French even though it is, now, and was built with local labor, then.

    Anyhow, I geography lots, the question is “when?”

  14. cafebabe says

    Back in the 1960s I worked in a nuclear research lab with much of this time spent on radiocarbon dating using the then best-practice gas counting method. Shielding the counters from cosmic rays was the holy grail that we pursued with fanatical fervour. At my lab we used about 30cm of mild steel and an annular tank with a couple of centimetres of mercury. Other labs used lots of lead, but modern lead is radiogenic and adds unwanted background to the counters. Medieval lead was made from different, non-radiogenic ores, and thus so valuable for this niche application that there was, according to legend, a thriving black marked for lead stolen from the roofs of medieval churches. It wasn’t the age of the lead that mattered, given the half-lives of the various steps in the decay chains, but the technology of the age in which it was made.

  15. John Morales says


    modern lead is radiogenic and adds unwanted background to the counters

    If it’s a known amount, then it can be subtracted from the signal, no?

  16. says

    @John Morales #18

    If it’s a known amount, then it can be subtracted from the signal, no?

    Not really for a number of reasons.
    1. The amount of background from your shielding is not known a-priori. You could measure it with the instrument you are shielding; However that leads us to point 2
    2. Subtracting the background is problematic because it increases the size of your error bar. Since it is “counting statistics” the error in a count rate is approximately the square root of the count rate. So, if you have a background signal B and a total signal T with Error_B = Sqrt(B) and ErrorT = Sqrt(T) then your foreground signal, F = T – B and Error_F ~ SQRT(T + B) (since errors are added in quadrature in this situation.) So if F is small and B is large you can easily get situations were the error is larger than the value it is an error for.
    3. Related to number 2. Even if you have excellent separation, if your signal is expected to be, on average, 1 count per second, and the intrinsic background from your shielding is, on average, 20 counts per second; how can you tell the difference between seeing your signal and seeing a statistical fluctuation in your background?
    It is mostly because of numbers 2 & 3 that we want to reduce background as much as possible because if we don’t the error will be large even if we subtract it. There are situations where you can make it work anyways, but those are significantly less common and require more sophisticated equipment and more than a basic detection signature (which for a given phenomenon nature does not always kindly provide us with.)

  17. says

    @cafebabe #17

    Medieval lead was made from different, non-radiogenic ores

    As I learned it… not quite.
    As long as lead has been mined the principle ore for lead has been Galena. So the ores it was mined from then are really no more or less radigenic than they are now. Perhaps in one region the ore might be mixed with granite (relatively rich in Uranium and Thorium) and in another it is mixed with limestone (fairly poor in said elements,) but that matters less. During the smelting process, even in ancient times, the long lived progenitors of natural radioactivity were mostly stripped out of the lead leaving it with just the relatively short lived activity. Since this process happened 500 to 1500 years ago the short-lived stuff that was still in the lead post smelting has had quite some time to decay out.
    In addition to not having the time to decay out, lead mined and smelted in modern times is contaminated with fallout from nuclear testing. It isn’t much; but it is quite noticeable because the isotopes in question are frequently “loud.”
    Finally, both medieval lead and modern lead are also subject to build of of isotopes produced by cosmic ray bombardment post smelting. This is what makes lead from shipwrecks, especially old shipwrecks (Roman shipwrecks being a common source) even better. That lead has had similar amounts of time to decay out as medieval and has been somewhat shielded from further activation by cosmic rays.

  18. cafebabe says

    What you said. Thanks for replying while us folks down under were asleep.
    As a matter of fact we kept a close track of the background, by once a week doing an overnight count on a sample prepared from coal. This would be rather older than the ~50k years range of carbon dating at the time. The problem with this background count is that its statistical error bars add inextricably to the putative errors in the sample you are trying to count.

    Occasionally, for a very old artefact, we would come up with a count which was lower than that of our coal standard. What can you say to the poor archaeologist? I once tried to suggest that if the client gave me a prior age guess I would supply a Bayesian probability distribution of age. This approach was frowned upon by my colleagues for perhaps obvious reasons.

  19. says

    No patterned damascus steel can hold a candle to anything made from mono steel properly chosen for the task at hand, or perhaps a san-mai/ni-mai from mono steels.

    That’s such a carefully qualified, yet broad, statement that it’s almost meaningless. “Properly chosen” monosteel doesn’t take into account the vast number of trade-offs hidden underneath it. Is 1095 or D2 always better than ${whatever} – no, not if cost is part of your criteria. And then you throw in layered architectures which, I remind you, are exactly the kind of think that you might get with a patterned damascus. In my posting, I deliberately referenced things like saw blades with deposited carbide on the teeth: it’s not a monosteel nor is it san/ni-mai, but the steel is doing nothing but serving as a heat sink and a carrier for carbide. I don’t even know (or care) what metal the disc of my diamond cut-off wheel is made of; it’s just a carrying surface and a heat sink.

    All of that brings me to an important point you appear to be overlooking: composite blades allow the designer to determine where they want strength, or flexibility, or low cost, or rust resistance, and what kind of properties they want on the cutting edge. You referred to san mai/ni mai, but that’s only two of the possible architectures for a blade. I could produce a blade of wrought iron with a cutting edge of pure carbide, if I wanted, by brazing it in there and using a diamond lapidary cutter to sharpen it. That would be stronger in pure terms than most monosteels, and sharper and tougher, as well. What it would not be is easier or cheaper.

    A blade of pattern-welded damascus (such as my suminigashi) that is layered around 1095 is going to be less prone to bending or breaking than a blade of pure 1095, and will have exactly the same edge properties.

    Yes, there’s damascus and tamahagane that are just done for aesthetic reasons, but there’s a lot more going on there than just “it’s pretty” – that said, I could characterize your comment as a critique of the massively tiled and twisted damascus patterns we often see (which you will not see in any of my blades) for aesthetic reasons. In those cases, the blades are as strong as the 1080 and 15N20 they’re made of.

    I’m also reluctant to treat monosteels as monolithic. A differentially quenched piece of 1095 has very different properties from a piece of 1095 that is purely martensitic. So, do you still want to consider it as a monosteel at all? What about something crazy like an L7 blade that was quenched at the bainite stage? It’s not anything like monosteel and it has unique properties.

    My preference is to think of blades as a combination of cutting edge (with certain properties) and the supporting metal behind the cutting edge (that has other certain properties) and perhaps facings on that supporting metal. In the case of a 1095 blade differentially quenched, I would think of that as a martensitic 1095 edge carried by a pearlite/bainite 1095 spine. Put a layer of 304 stainless steel on either side of the spine and you can still talk about the different properties you expect in the blade rather than just calling it a san mai or whatever. And, it is possible to produce things that are better in many ways that monosteel. For example, as I write this, I am looking at a differentially quenched (edge quench) blade with a 1095 edge at RC60, a core of 1095 at RC30, and facings of 304 stainless that are soft. I can’t just talk about the cutting edge (it’s sharp!) without considering that the core is basically a spring and the sides are corrosion-proof. That’s an example of why I decompile my description of blades into edge/carrier/facings rather than just “pattern damascus” or “monosteel.” If you just work with monosteel you can avoid having to have a terminology that captures the complexity.

    And aesthetics are something else, entirely.

    PS- last week “This Old Tony” dropped a video of him using TIG filler rod that is tool steel. He cut an axe-shape out of (?) on his CNC mill, then built up an edge-bead of tool steel with the TIG, sharpened it, and it’s tough and sharp. He also took some mild steel, built an edge with the filler rod, sharpened it, and used it as a cold chisel to cut another piece of mild steel apart. That’s an example of another case where the terminology of edge/carrier/facings is sufficiently descriptive.

  20. Owlmirror says

    Today’s Syria was part of Persia in the 11th century, which is the time I am talking about – when Damascus steel was a big deal.

    WikiP:Damascus steel says that the production centers were in Khorosan and Merv; further down it also says Isfahan (all locales in Persia/Iran proper).

    In fact, the Wikipedia article suggests that it is at least possible that Damascus steel is not named specifically for the swords made in the city of Damascus. Huh. I wonder if it might be analogous to the tuxedo — lots of places were making that kind of expensive fancy suit; Tuxedo Park just got lucky in associating that product with that name because of the Tuxedo Club.

    Wait. Back to geography, and also history:

    Syria was part of Persia in the 11th century

    I’m pretty sure that Syria was not part of Persia in that time period.

    Are you confusing the Abbasid Caliphate with the Sassanid Empire?

    the Syrians would not have been trading and incorporating steels from India on their own, that was a result of being part of Persia

    Trade routes aren’t a thing? It seems a lot more odd to me to think that only Damascus swordsmiths were using steels from India, as opposed to swordsmiths all along the trade routes from India outward.

Leave a Reply