Forging

Safety

Carbon monoxide is what actually kills smiths — a gas forge in a shut shop, a coal fire with no draught. Alarm on the wall, door open. Hammer on anvil peaks at 110–120 dB and hearing does not grow back: plugs from the first day. Never heat galvanised, plated or unknown scrap — zinc buys you a night of fever, cadmium can kill, and sulfur-bearing free-machining steel crumbles at forging heat. Side-shielded glasses always; scale comes off the anvil as well as the work. Cotton, wool or leather, never synthetics, and boots you can kick a hot offcut out of. A glove on the tong hand only — the hammer hand needs to feel the blow. Hot steel and cold steel are the same colour. Check the hammer's wedge, never leave a hardy standing in the hole, and wet the fire down and sit with it a while before you lock up.

Done When

Sighted down its length it is straight, and it rocks nowhere on the anvil face; the section calipers what it was drawn to. Wire-brushed and read in raking light there is no dark seam — a suspect line opens under the corner of a chisel if it is a cold shut, and does nothing if it is only scale. The shape came from metal moving, not from grinding it away. And a coupon off the same bar, normalised and snapped, breaks to a fine silky grey: bright, coarse and faceted means it was too hot for too long, which is a verdict on the heat, not on the hammering.

Steps

Pick the steel and its window. Mild works from 1,250 °C down to a dull red and forgives most things; high-carbon and alloy steels want 1,150 down to about 850 °C and no lower; air-hardening tool steel goes into vermiculite straight off the anvil, or it hardens overnight and cracks in the bin.

Run the fire reducing, not oxidising — the pocket above the tuyere under a cover of coke, or a gas burner tuned slightly rich. An oxidising fire takes a percent of the stock as scale every heat, and pulls carbon out of the surface while it does it.

Judge colour in shade, never in sunlight: the eye reads heat only against its surroundings, and a bar that looks dull outdoors is already burning. Yellow ≈ 1,150 °C, orange ≈ 1,000, cherry ≈ 800 — and a magnet letting go is 770 °C exactly, the one fixed point you can check by hand.

Soak, do not just heat: about a minute per 5 mm of thickness once it is up to colour. A hot skin over a cold core tears when you move it.

Work the heat, then stop. At 1,200 °C the bar radiates some 250 kW/m², and a 12 mm section sheds about 15 °C a second — the thirty seconds everyone quotes. Twice the section is twice the time; a 5 mm taper is done in under ten; and thin work left lying on a cold anvil face loses more by conduction than by air.

Draw out over the far edge of the anvil or under a fuller, half-on and half-off, turning 90° each pass: a narrow contact drives metal along the bar instead of sideways. Planish flat afterwards with light, level blows.

Never draw a round as a round. Take it to square, draw it there, knock the corners to an octagon, and round the octagon last — a cylinder hammered across its diameter turn after turn puts its own axis into tension and opens a hole down the middle.

Upset with no more than two and a half diameters of hot bar standing unsupported, and straighten every few blows: friction pins the ends while the middle spreads, so it barrels first and folds second.

Punch from both sides, not through: sink halfway, flip onto the raised mark, drive the slug out. Withdraw the punch every few seconds and cool it, and drop a pinch of coal dust in the hole so the steam pops it free. Punching bulges the bar — dress it back; drilling would cut the grain instead of parting it.

Brush the scale off before any blow that matters. The oxide does not move like the metal under it: a blow presses it in as a pit, or folds it in as an inclusion that can never weld shut.

Finish low — the last blows just above A₃ — so the strain you have only just put in recrystallises into fine grains instead of feeding coarse ones. Then normalise: three descending air cools, about 900, 855 and 815 °C, each soak shorter than the last.

Tools

An anvil of 50 kg or more with 70–90 % rebound — drop a 25 mm ball bearing from 250 mm and measure — bedded solid at knuckle height; it wants to outweigh the hammer about fifty to one, or the blow's energy goes into moving the anvil instead of the steel. A 1–1.5 kg cross-pein or rounding hammer (velocity is squared in ½mv², so speed beats mass, and a hammer too heavy to swing fast is a slower hammer), and a 3 kg sledge if someone will strike for you. Tongs that fit the section, with a ring to close them. Hot cut, fullers, a butcher, punches and drifts with a can of water beside them, a monkey tool, a swage block. Coal or coke under a hood with a blower, or a gas forge whose burner can be tuned rich. Steel wire brush, a magnet, calipers, a bucket of vermiculite, and a slack tub for the tools, not the work.

The Science

Hot working is not about softness, it is about recrystallisation. Above roughly 0.4 of iron's melting point in kelvin, new strain-free grains nucleate as fast as the hammer distorts the old ones, so the steel never work-hardens and can be moved as far as you like; the stress needed to move it falls by an order of magnitude between cold and 1,200 °C. The floor is not the recrystallisation temperature but the A₃ line — 912 °C at zero carbon, falling to 727 °C at 0.77 % — below which ferrite appears beside austenite and the two shear apart. The ceiling is set three times over: austenite grain coarsens above about 1,050 °C; sulfur-bearing steel goes hot-short at the 988 °C Fe–FeS eutectic; and near 1,350 °C the grain boundaries oxidise and the steel is simply burnt. Temperature and speed trade against each other through the Zener–Hollomon parameter Z = ε̇·exp(Q/RT), Q ≈ 300 kJ/mol in austenite: hitting ten times faster is worth about 130 °C of cooling — which is why a slow press reaches the centre of a billet and a fast hammer only moves its skin. The metal flows perpendicular to the hammer face toward the nearest free surface, and the ratio of contact length to section thickness decides how deep that reaches: near 1 it goes through, below about 0.5 the surface moves while the axis is pulled into tension. The grain flow then follows the finished shape instead of being cut across it, which is what makes a forged part stronger than a machined one along the flow — and weakest across it.

How It Fails

Cold shut: a lap hammered closed over its own oxide. It is not a weld and never becomes one, and by Griffith's criterion the stress needed to run it as a crack falls as one over the square root of its length — so it is the flaw that fails the part years later. Central burst, or an end that opens into a fish mouth, from drawing a round as a round or from bites too shallow for the section: the skin moves and the axis is pulled apart. Hot shortness — edges crumbling like wet chalk at yellow — from sulfur, or from copper picked up out of scrap; both melt into the grain boundaries below forging heat. Cracks from working under the A₃ line, or from hammering through the 250–400 °C blue-brittle range where nitrogen pins the dislocations. Coarse grain from soaking at yellow, which normalising will undo; burnt steel, which sparks and crumbles and nothing will undo — cut it off. A decarburised skin that will not harden afterwards, from too many long oxidising heats: forge a blade oversize and grind through it.

Summary

Moving hot steel with a hammer — drawing out, upsetting, bending, punching, spreading — inside the window where it flows freely: above the temperature at which it cracks, below the one at which it is ruined.