Smiths were doing all of this for roughly three thousand years before anyone could say why it worked. They knew that a red bar goes soft, that a quenched bar goes hard, that a blued bar stops snapping. They knew it the way you know how to catch a ball. What they did not have — could not have — was a look inside.
When somebody finally looked, in a private laboratory in a Sheffield house in the 1860s, the answer turned out to be simpler than anyone expected, and it explains the whole craft at once. This course is built on that one look. We will not begin with how to swing a hammer. We begin by asking what metal actually is — and once you have that, annealing, hardening, tempering and the shape of a blade stop being four separate lessons and become the same lesson told four times.
You have handled metal since before you could read. A spoon, a key, a nail, a coin. It is the most familiar material in your life and it behaves like nothing else you own, and almost nobody ever tells you why.
Take a paperclip. Bend it. Now bend it back. Bend it back and forth a dozen times and two odd things happen: it becomes noticeably stiffer at the bend, harder to move than it was at the start — and then it snaps. You did not add anything. You did not remove anything. You bent a piece of wire and the wire changed its mind about what kind of thing it was.
Now a harder question. A blacksmith and a file-maker can be handed identical bars from the same batch of steel. One turns his into a spring that will bend double and come back, forever. The other turns his into a file so hard it will scratch that spring, and so brittle that dropping it on a stone floor breaks it in two. Same bar. Same metal. Same everything. The whole difference is what was done to it with heat and hammer.
So something inside metal is being changed, invisibly, by things that look from the outside like nothing at all — waiting, glowing, cooling fast, cooling slowly. Before you can work metal with any confidence, you need to know what that something is.
Here is the sentence this whole course rests on. It sounds wrong the first time, because metal is the smoothest, most seamless, most obviously solid stuff you will ever hold:
Not one crystal. A crowd of them, jammed shoulder to shoulder, nearly all too small to see.
Start with what a crystal is, in plain words: atoms stacked in a regular repeating order, like tins stacked in rows on a shelf, rows on rows on rows, all the same way up. Salt does this. Diamond does this. Metal does this too — every metal, all the time, including the spoon in your hand.
The crowd part comes from how metal freezes. When molten metal cools it does not solidify as one tidy stack starting at one corner. It begins freezing in thousands of separate places at once, and each of those places starts its own stack at whatever angle it happens to start at. Each little stack grows outward until it runs into its neighbours, and there it stops — because where two stacks meet, their rows are at different angles and cannot be persuaded to line up. You get a ragged seam.
Each of those little stacks is a grain — one crystal, one consistent direction of stacking. The ragged seams between them are grain boundaries. A finger-thick bar of steel holds many millions of grains, and the whole of metalwork is the business of changing their size, their shape, and how badly tangled they are.
Hold onto this, because it reframes every problem you will ever have at an anvil. A bar that has gone soft is not tired. A bar that has gone hard is not angry. In both cases the crowd has been rearranged, and your entire job — hammer, fire, water, oil, time — is arranging it on purpose instead of by accident.
Stone is made of crystals too. Glass is not — glass has no repeating order anywhere in it, which is precisely what makes it glass. But hit either with a hammer and it breaks. Hit metal with a hammer and it moves. That difference is the reason there is a craft here at all, and it comes straight out of the crowd.
So being crystalline is not what does it. The second half of the answer is how metal atoms are held together. In stone and in salt each atom is bonded to particular neighbours in particular directions, and those bonds must be broken to move anything. In a metal the atoms give up their outermost electrons into a shared pool that runs through the whole piece, and every atom is held by that pool rather than by any named partner. The bond does not care which neighbour an atom sits beside. Slide a whole layer along by one atom and nothing has been broken — every atom simply has a new neighbour, and the pool holds it just as tightly as before. That same shared pool of loose electrons is also why metal conducts heat and electricity and why it is shiny. One fact, three familiar consequences.
Inside a grain, the atoms sit in neat layers. Those layers can slide over one another — the trade word is slip, meaning one plane of atoms shifting bodily across the plane beneath it. That is what bending is. Nothing breaks; a stack of layers simply shears a little, like a deck of cards pushed sideways.
But the layers do not slide all at once. Shifting an entire layer in one go would take an enormous push — every atom in it would have to climb out of its seat at the same instant. That is not how it happens, and the reason is the most elegant thing in this course.
Think about moving a heavy rug across a floor. Dragging the whole rug at once is brutal work, so nobody does that. You kick a ruck — a wrinkle — into one end and walk the wrinkle across to the other side, and when it comes off the far edge the whole rug has moved, an inch at a time, for almost no effort. You never moved the rug. You moved a wrinkle.
Metal does exactly this. Inside every grain there are flaws in the stacking — places where one row of atoms ends early, leaving a step, a wrinkle in the crystal. Metallurgists call such a flaw a dislocation, which just means a place where the stacking is out of place. Push on the metal and the wrinkle travels, one row of atoms at a time, and when it reaches the far side of the grain the whole layer above has shifted by exactly one row.
This is why metal is ductile — able to be drawn out and reshaped without breaking — while stone is not. Stone's bonds point in fixed directions, so there is no plane its atoms can slide along without tearing bonds apart, and a push has nowhere to go but into a crack. Glass, with no ordered layers at all, has nowhere to run a wrinkle either. And it is why metal is worth having: it is the only common material you can hit repeatedly and reshape rather than destroy.
A wrinkle runs freely inside a grain, because inside a grain the rows all point the same way and the road is straight. Then it reaches the boundary — the seam where the next grain's rows sit at a completely different angle — and the road stops. The wrinkle cannot simply carry on into a stack that is facing elsewhere.
Follow that through, because it hands you a rule you will use for the rest of your life:
You can see coarse grain without a microscope, once you know the tell. Bend a piece of metal that has been overheated and the surface goes lumpy and matte, like the skin of an orange. Each lump is one enormous grain shifting on its own. Sheet-metal workers call it orange peel, and it is a visible confession that the crowd got too big.
Which sets up the central tension of the whole craft. Heat is what lets you shape metal — and heat is also what grows the grains too big. You cannot have one without risking the other. Managing that is the trade.
Come back to this as you go. Right now only the first picture will mean much. By the end of Part Five all five will, and you will be able to say what each one is good for without being told.
Heat, at bottom, is atoms jiggling. Cold atoms jiggle a little and stay in their seats. Hot atoms jiggle hard. Heat metal enough and the jiggling becomes violent enough that atoms can leave their seats altogether and find better ones.
Now watch what that does to a crowd that has been mangled by hammering. Two things happen, in order, and it is worth knowing they are two things and not one.
First, at a modest heat, the wrinkles shuffle themselves into tidier lines and some cancel each other out. The metal eases. Metallurgists call this recovery — the stress comes out, but the grains keep their battered shapes.
Then, higher, the real event. Brand new grains sprout inside the old mangled ones — small, clean, unstrained, starting from scratch — and they grow outward and eat the damaged material completely. This is recrystallisation, which means exactly what it says: the crystal crowd is remade. It begins at roughly a third to a half of the metal's melting temperature — counted on the absolute scale, from absolute zero rather than from room temperature — which is why a low-melting metal like lead recrystallises while sitting on the bench, copper wants a dull red, and steel wants a great deal more.
What comes out the other side is soft, fine-grained and completely willing. Every wrinkle from every blow you struck is simply gone, because the material those wrinkles lived in no longer exists. That process has a name you already know: annealing.
And immediately the danger. Keep holding it hot after recrystallisation has finished and the new grains have nothing left to eat but each other. The big ones swallow the small ones. This is grain growth, and it is how a perfectly good piece gets quietly ruined by being left in the fire while you answer the door. Anneal hot enough and long enough, and no longer.
One honest complication, because it will bite you otherwise. The word "anneal" covers two different jobs:
Now put the last two ideas side by side. Hammering tangles the crowd. Heat remakes the crowd. What happens if you do both at once?
That is forging. Above the recrystallisation temperature the crowd re-forms as fast as your hammer can tangle it. The damage from each blow is erased before the next blow lands. The metal never gets harder, never gets tired, never reaches a limit — you can move it all day and it stays as soft and willing as when you started.
This is the whole reason a smith works hot rather than cold, and it explains the one piece of forge discipline every beginner is given and few are given the reason for: stop when the colour dies. Above a good working red, every blow is shape. Below it the remaking has stopped, and the identical blow is now doing damage instead — jamming the crowd, and eventually splitting it. The hammer did not change. The metal did.
Two related cautions worth carrying:
You will notice nobody at a forge is holding a thermometer. They are reading colour, and colour is not a rough guide — it is a genuine, repeatable report of temperature, good to within a few tens of degrees in the hands of somebody who has looked at it for a season.
Everything hot glows, and what it glows is set by how hot it is, not by what it is made of. This is the same physics as the fire course's yellow flame, which glows because it is full of solid carbon specks too hot to stay dark. Steel at 700 degrees and a soot speck at 700 degrees are the same colour, because the colour reports temperature and nothing else.
So when you read the bar you are not reading some rough stand-in for what the crowd is doing. You are reading the one thing that decides what the crowd is doing.
Two working habits fall straight out of this, and neither is fussiness:
Go back to the paperclip. Bend it cold, and wrinkles start running through the grains. But here is the part that makes the whole thing click: moving wrinkles create more wrinkles. Deformation multiplies them enormously — a heavily hammered piece can hold hundreds of thousands of times as many as an annealed one.
And wrinkles get in each other's way. Two of them crossing paths lock together. A crowd of them tangles like a knot of rope. Each new wrinkle finds the road already congested with the wrecks of the last ones.
So the metal becomes harder to move the more you move it. That is the stiffening you felt in the paperclip, and it has a name: work hardening, also called cold working — shaping metal below the temperature at which the crowd can remake itself.
This is not a defect. For thousands of years it was the only hardening anybody had. Copper and bronze cannot be hardened by quenching — nothing you do with water will help them — so the whole Bronze Age hardened its edges by hammering them cold, and a cold-hammered bronze edge is genuinely, usefully hard. The technique never went away: European scythes are still sharpened by peening, hammering the very edge thin and hard on a small anvil, with no fire involved at all.
You are using it constantly without noticing. The rim of a copper pot is stiffened by hammering. A silversmith planishes a bowl with hundreds of light overlapping blows and the bowl goes rigid under his hands. Drawn wire is stiffer than the rod it came from, because being hauled cold through a draw plate — a hard block with a tapered hole a shade smaller than the rod — is itself a cold working. Rolled sheet is stiffer than cast metal of the same alloy.
Keep going and the traffic jam becomes total. Every road blocked. No wrinkle anywhere in the piece can move another step.
Now strike it again. The blow has to go somewhere, and the one form of movement still available to a crowd that cannot slip is to come apart. A crack opens, usually along a grain boundary, and runs.
That is the snap at the end of the paperclip. Not metal fatigue in the strict sense — that is a slower, sneakier cousin that kills things under repeated small loads over thousands of cycles — but the plain end of the road. You spent the metal's whole capacity to move, and then asked for more.
Which gives you the single most useful working rule in cold metalwork:
In practice: raising a copper bowl from flat sheet may take six or eight annealings on the way. Bending a stubborn decorative scroll cold, you anneal when it starts to fight. Setting a rivet, you get a few blows and should not be tapping at it endlessly. If the metal has begun to resist you noticeably, it is not being awkward — it is telling you the roads are full.
And its opposite, which will save you a blade one day: never cold-straighten a hardened piece. Hardened steel has almost no budget at all — Part Four explains why — and the "gentle" bend that would correct a soft bar simply breaks it.
Put Modules 5, 8 and 9 together and you find you already own a complete control system, before any mention of quenching, and it works on every metal there is.
Notice that hardness and softness are not properties the metal has. They are positions on a scale that you put it in, and can take it out of again with fire. Sit with that, because Part Four is about to do the same trick by a completely different mechanism, and it should feel familiar rather than strange.
Everything so far applies to all metal. What follows applies to steel alone, and the reason is a tiny amount of carbon.
Pure iron is soft. Genuinely soft — it marks easily, it bends easily, and no amount of heating and quenching will make it hard. Wrought iron, the old structural material, is near enough this: lovely to forge, slow to rust, and useless for an edge.
Add carbon and everything changes. An alloy — a metal with something else deliberately mixed in — of iron with a little carbon is steel, and the amounts involved are startlingly small:
Consider how strange that is from outside. One part in a hundred of a black powder decides whether you have a bracket, a razor or a frying pan. And for most of history nobody could measure it. Carbon got into iron by accident, from the charcoal it was smelted with, and generations of smiths learned to judge it by how a bar sparked on a grindstone, how it moved under the hammer, and how it behaved when quenched. Whole traditions were elaborate ways of getting a controlled amount of carbon into iron without ever knowing that carbon was the thing being controlled. Sheffield packed bars of iron in charcoal and baked them for a week, so that carbon crept in from the outside and left the bars covered in blisters. India and Persia sealed iron and charcoal together in a clay pot and melted the lot, so the carbon mixed in evenly — the crucible steel that western smiths spent centuries failing to copy.
Now the fact that makes steel steel. Iron has two different ways of stacking its atoms, and it swaps between them at a particular temperature. What matters is not how tightly the iron atoms sit — it is the size of the gaps left between them, because that is where a carbon atom has to live.
Cold, iron stacks in an open, boxy arrangement, and counter-intuitively that open arrangement leaves only small, awkward gaps: lots of them, all too cramped for a carbon atom. So the carbon is thrown out and gathers separately, as a hard brittle compound of iron and carbon called iron carbide, laid down in fine layers among the plain iron. In slowly cooled steel those layers are so fine and regular that under a microscope they shimmer like mother-of-pearl, which is how the structure got the name pearlite. The plain iron grains beside it are called ferrite. Soft, stable, ordinary steel.
Heat past roughly 723 degrees Celsius (about 1,330 Fahrenheit) — the exact figure moves a little with the carbon content — and the iron re-stacks. The new arrangement packs the iron atoms closer together, and yet it leaves fewer and far bigger holes between them, the way stacked oranges leave a few roomy hollows while a stack of boxes leaves many mean ones. Those bigger holes can hold carbon, dissolved evenly right through the metal, like sugar in tea. That hot arrangement is called austenite, after Sir William Chandler Roberts-Austen, an English metallurgist of the late nineteenth century. The temperature at which the swap happens is the critical point, and everything about hardening steel happens above it.
From here on this course calls them the roomy hot packing and the tight cold packing. Remember that the words describe the gaps, not the iron — it is the gaps that carbon cares about, and carbon is what the rest of Part Four is about.
So to harden steel you must first get it above the critical point and hold it there long enough for the carbon to spread out evenly. In practice that means a little above — a good even cherry to orange judged in shade, held briefly, no longer than needed.
The same critical point gives you the other thing you should do before hardening, and which beginners always skip. Normalising: heat just past critical, take the piece out of the fire, and let it cool in still air until the glow is gone. Do that two or three times. Each cycle recrystallises the steel, and each cycle leaves the grain finer than the last — undoing the coarse grain and the uneven stresses your forging put in. Normalise before you harden and you will have fewer cracks, fewer warps and a tougher piece, every time.
Here is the whole of hardening, and it follows from Module 12 with no new ideas at all.
Your steel is above critical. The iron is in its roomy packing, carbon dissolved evenly throughout. Now it has to come back down, and the iron badly wants to return to its tight cold packing. The only question is what happens to the carbon while it does.
Cool slowly and the carbon has all the time it needs to walk out of the way and gather into its layers. The iron settles comfortably into its tight packing around it. You get soft, ordinary, pearly steel — which is exactly what annealing steel means, and why it must be cooled slowly.
Cool very fast and the carbon has nowhere to go and no time to go there. The iron snaps back toward its tight packing regardless, but the carbon atoms are still sitting inside it, and the tight packing has no room for them. Every trapped carbon atom wedges the stacking apart. The crystal cannot form properly, so it forms wrong: strained, distorted, in fine needle-like plates rammed together at angles.
That trapped, strained, wedged-open structure is called martensite, after Adolf Martens, a German metallurgist of the same period. It is the hardest thing plain steel can be, and it is hard for a reason you now already understand: a crowd that strained, that fine, and that full of obstacles has no road left for a wrinkle to run down. Nothing can slip. Nothing can bend.
And there is the price, in the same sentence. Metal that cannot slip cannot bend. Freshly quenched steel is glass: it will scratch almost anything, and it will break if you look at it unkindly. Left on the bench overnight it can crack all by itself, from the strain it is holding.
Three practical consequences worth carrying to the tank:
A freshly quenched blade is useless. It is as hard as you will ever get it, and it will snap in the first hard cut. So the very last thing you do is deliberately make it softer — and understanding why turns tempering from a mysterious extra step into the obvious next move.
First, two words that get used as if they meant the same thing, and do not:
They pull against each other, and the reason is exactly the crowd. Hardness comes from a jammed crowd that cannot slip. Toughness comes from a crowd that can slip a little, and so can soak up a blow by moving instead of splitting. You cannot have both at their maximum, in the same piece of steel, at the same time. Ever. The craft is choosing where between them to stand.
Tempering is how you choose. Take the quenched piece and reheat it gently — somewhere between about 150 and 350 degrees Celsius (roughly 300 to 660 Fahrenheit), nowhere near a glow — and hold it there a while. That is warm enough for the trapped carbon to creep a very short distance and settle into tiny particles, which lets the wedged-open stacking relax back toward where it wanted to be.
Strain comes out. Hardness goes down. Toughness comes up. And how far all three of those go is set by one thing: how hot you let it get.
In Module 7 you read the metal's own glow. At tempering heats the metal is far too cool to glow at all — and yet a smith with no thermometer can hit two hundred and twenty degrees on a chisel edge, reliably, by eye. He is reading something else entirely.
Polish a piece of steel bright and warm it. Oxygen in the air grows a film of oxide on the clean surface, and the hotter it gets the thicker that film grows. While it is still extraordinarily thin — far thinner than a hair — it does something lovely: light bouncing off the top of the film and light bouncing off the steel underneath interfere with one another, and certain colours cancel while others survive. The surface takes on a colour that depends on nothing but the film's thickness.
This is the same effect as the rainbow in a puddle with a drop of oil on it, and the colours in a soap bubble. And because thickness reports temperature, the colour reports temperature. It is one of the loveliest pieces of applied physics in any craft, and it was in daily use for centuries before anyone could explain it.
Switch the second demonstration above to Tempering colours and walk through the sequence. Pale straw first, then darker straw and bronze, then brown-purple, purple, dark blue, light blue, and finally grey — at which point you have gone too far for any edge tool.
Two honest limits. The colour only appears on clean, bright steel — a sooty or scaled surface tells you nothing. And the film is a one-time report: once grown, it stays. You cannot heat a piece that is already blue and read it again, because you would be looking at yesterday's answer. Sand it bright, or use a thermometer.
Now the derivation reaches the thing that makes swordsmiths sound like mystics, and it turns out to be ordinary bookkeeping.
A blade wants two contradictory things in two different places. The edge wants maximum hardness: it must not dent, must not roll, must hold a fine geometry against wood, bone and rope. The spine wants toughness: when the blade is levered or struck sideways it must bend and come back rather than snapping in two.
Module 14 said you cannot have both at once. That was true — of one piece of steel treated one way. But nothing stops you treating one end differently from the other. The crowd is local. It can be arranged differently an inch apart.
Three ways to do it, all in use today, all built on nothing you have not already read:
And this logic goes everywhere once you have it. An axe wants a hard bit — the cutting end — and a soft eye, the hole the handle passes through, so the eye can take a handle and a wedge driven into it without cracking. A hammer wants a hard face and a soft middle — the soft middle is what stops the whole head splitting when the face takes a bad blow. A file wants to be hard everywhere and is accordingly brittle everywhere, which is why files break rather than bend, and why they make such tempting and such dangerous raw material for a beginner's knife. A spring wants toughness everywhere and hardness nowhere in particular.
Ask, of any tool: which part of this must not dent, and which part must not crack? The answer tells you the treatment.
There is a sequence, and each step is there because the step before it left something that needs answering. Nothing in it is ritual.
Before you diagnose anything, settle one question first, because half the failures below are really this one. Draw a sharp file across the quenched piece. If it skates and squeals, the piece hardened; if it bites and cuts, it did not. A crowd that cannot slip cannot be cut into — so the file is reading the crowd directly. That is a complete, free, thirty-second answer, and it will stop you tempering a piece that never hardened in the first place.
And here is the part that turns this course into a working skill. Every common failure at the anvil is a plain statement about the crowd, and you can now read all of them:
If you followed all of that, you were not handed a list of techniques. You were handed one fact and its consequences — and it is worth laying the chain out end to end, so you can see that nothing else was ever smuggled in.
Metal is a crowd of tiny crystals.
Therefore bending is wrinkles running through the crowd, not atoms being torn apart — which is why metal reshapes where stone shatters. Therefore boundaries between grains stop wrinkles, so fine grain is strong and coarse grain is weak. Therefore heat, which lets atoms leave their seats, allows the crowd to be remade from scratch — that is annealing, and it makes metal soft and workable again. Therefore holding it hot too long lets big grains eat small ones, and you can ruin a piece simply by waiting. Therefore hammering above the remaking temperature never hardens anything, because you are hammering and annealing in the same instant — which is why the smith works hot and stops when the colour dies.
Therefore hammering cold multiplies wrinkles and tangles them, which makes metal harder and stiffer — until the jam is total, at which point the next blow has nowhere to go but into a crack. Therefore cold work is a budget, and annealing refills it. Therefore the Bronze Age could harden an edge with nothing but a hammer.
Therefore, in steel, a little carbon rides inside the hot roomy packing of iron and must leave when the iron goes back to its tight cold packing — so cooling slowly lets it leave, giving soft steel, and cooling fast traps it, wedging the stacking open and freezing the crowd mid-rearrangement. Therefore quenched steel is the hardest and most brittle thing plain steel can be, because a crowd that cannot slip cannot bend. Therefore gently reheating it lets just enough strain out — that is tempering, and it is a deliberate trade of hardness for toughness at a price you set with the temperature. Therefore a blade, which needs a hard edge and a tough spine, gets two different crowds in one bar — by clay, by edge quench, or by drawn colours. And therefore temperature is the one dial that sets all of it — which is why a smith spends his life reading colour. Colour does not report the crowd directly; it reports temperature, honestly and repeatably, and temperature is what the crowd obeys. Two different colour languages do it: the metal's own glow at forging heat, and a film of oxide at tempering heat.
And therefore you can now answer questions this course never asked. Why a cast-iron pan cracks rather than dents. Why an old spring makes a better beginner's knife than an old file. Why a hardened blade must be straightened hot. Why a maker normalises three times before a quench and looks, to an onlooker, like he is wasting fuel. Why the mark on a Japanese sword is a map rather than an ornament. Why bending a wire coathanger back and forth heats it, stiffens it, and then breaks it at the same spot every time.
That is what a first-principles education buys. Not more answers. Better questions, and the ability to derive the answers yourself when the fire is lit, the bar is glowing, and there is nobody in the shop to ask.
"Iron sharpeneth iron; so a man sharpeneth the countenance of his friend."
— Proverbs 27:17, King James Version, 1611
The Guild of the Forge · Course One of the Twelve Guild Courses · Iron Roots Supply
Back to the Guild Hall · The fire course, The Chemical History of a Campfire, is the fire this one is built on.
You can also train Blacksmithing in the world of Iron Roots — seek the Master Smith at The Forge, and claim the guild's Smith's Book from him.
Old methods, modern backing, honestly explained.