Tamahagane forging: tatara furnace, folding steel, and yaki-ire quench

A finished katana looks like one piece of steel. Under the polish it is a negotiation between hard and soft metal that started days earlier in a clay furnace and ends, if luck holds, in a few seconds of water. This article stays upstream of the broader sword-making overview: how satetsu becomes a kera bloom, how smiths sort tamahagane, why they fold it, how high- and low-carbon layers are stacked, and what yaki-ire actually changes in the metal. Part names like ha, mune, and hamon live in sword anatomy; famous named blades that rode this process sit in famous swords.

Modern blast-furnace steel is designed to be uniform. Tatara bloom steel is designed to be sorted. That difference is the whole craft. If you only remember one idea, remember this: uneven carbon is not a defect the Japanese tradition failed to fix—it is the raw material the forge is built to manage.

Satetsu: iron sand, not iron ore bars

Traditional Japanese sword steel starts as satetsu(砂鉄)—iron-bearing sand that magnets pick up from riverbeds, coasts, and washed mountain soils. The San'in region, especially Okuizumo in Shimane Prefecture, became famous because granite country sheds usable sand into streams. Historical mining often used kanna-nagashi (鉄穴流し): cut a hillside, wash soil with water, and let dense iron particles settle. Those stepped wash landscapes still scar parts of the Chūgoku mountains.

Two sand families show up in smith talk. Masa (真砂) sand, linked to weathered granite, is the feed most often named for modern sword-grade kera-oshi runs at the Nittoho Tatara. Akome (赤目) sand, associated with other rock types, historically fed different smelt styles—including longer runs aimed more at pig iron (zuku). Popular summaries sometimes map masa to soft core stock and akome to hard jacket stock; working practice at the surviving furnace emphasizes masa for the bloom smiths grade into blade steel. Either way, the sand is not “pure iron.” It is oxide particles that need charcoal chemistry and human patience before anyone hammers a billet.

Charcoal does double duty: heat and carbon donor. Pine charcoal burns hot; other woods trade stability for peak temperature. Wet charcoal is sabotage—steam and temperature swings wreck reduction. Before a smelt, the clay furnace floor itself is dried for days or weeks so moisture does not explode the chemistry when the fire finally rises.

The tatara furnace: three days of controlled chaos

A tatarais a boxy clay furnace with bellows forcing air through the charge. The word can mean the furnace, the bellows action, or the whole worksite. Unlike a European blast furnace that runs for months on ore and coke, a traditional Japanese sword tatara is built for a campaign, then broken open. At today's Nittoho Tatara—revived in the late 1970s under the Society for Preservation of Japanese Art Swords (NBTHK)—a documented display bloom came from about 10 tons of iron sand and 12 tons of charcoal, yielding a kera on the order of 2–2.5 tons after roughly three continuous days and nights.

1. Day one — raise and stabilize. Sand and charcoal go in; bellows wake the fire. Early temperatures sit lower (often cited around 800–1000°C) while the bed dries and reduction begins.
2. Day two — peak reduction. Furnace interiors are reported near 1400–1500°C in peak descriptions. Iron oxides reduce; carbon migrates into metal; slag separates. The master smelter (murage) adjusts feed rate and air—on the order of tens of kilograms of sand per hour in published schedules—watching color, smell, and slag behavior more than a digital readout.
3. Day three — finish and break out. Feeds taper toward a maximum bloom mass. Walls come down. The black, slag-crusted kera is hauled out to cool before anyone trusts a hammer near it.

Only a fraction of that kera is first-grade tamahagane. Commercial explanations often put usable jewel steel around a quarter of bloom mass, with the rest lower grades, pig iron, and waste. Annual output from a few winter campaigns is measured in tons, not warehouses—enough to remind you why licensed smiths treat each parcel like rationed spice, not commodity bar stock.

Kera-wari: reading fracture faces

After cooling, workers smash the kera (kera-wari) and sort chunks by fracture color, density, sound, and impurity. High-carbon faces tend to look brighter and silvery; lower-carbon pieces read darker. Grades matter:

• Tamahagane suitable for blades—commonly discussed in the ~0.6–1.5% carbon window, with finer internal grades (first-, second-, third-class) used by suppliers and smiths.
• Zuku—very high carbon pig-like iron (often cited above ~1.75–2%), useful for casting or further work, not a finished sword billet.
• Soft / low-carbon pieces—stock for cores, spines, or other soft roles once forged clean.

This sorting table is where the “hard vs soft” story begins. A smith does not melt the whole bloom into one average. They choose piles the way a baker separates pastry flour from bread flour. Get the piles wrong and every later fold just mixes the mistake more thoroughly.

Process map from sand to quench

Orikaeshi-tanren: what folding actually fixes

Raw tamahagane still holds slag stringers, oxide pockets, and carbon patches. Orikaeshi-tanren (折返し鍛錬)—heat, flatten, fold, forge-weld, repeat—attacks those problems at once. Hammer blows squeeze slag toward the surface where it sheds with scale. Each weld interface multiplies layers and forces carbon to diffuse across former boundaries. The polished surface pattern later called hada (wood grain, straight grain, and so on) is a fossil of those welds, not a magic sharpness multiplier.

Layer math grows fast: one fold doubles, so ten folds are on the order of a thousand layers, fifteen folds tens of thousands—if every weld is perfect. Real billets lose material to scale and trimming, and welds are never textbook-clean. The useful question is not “how many layers?” but “is carbon even enough, and is slag gone enough, for this billet's job?”

Metallurgical work on hagane folding found that cycling near the traditional twelve to fifteen range can land edge steel around ~0.55–0.60% carbonwith quench hardness suitable for a cutting edge (~800 HV in the study's framing), with about thirteen cycles as a sweet spot in that dataset. Fold far past the useful band and you keep oxidizing carbon away—the billet gets “cleaner” in a trivial sense and worse as edge steel. That is why the million-fold sales pitch is backwards: folding is quality control with a diminishing return curve, not a scoreboard.

High-carbon vs low-carbon layers: kawagane and shingane

After separate billets are folded clean, smiths rebuild a composite bar (tsukurikomi). The classic beginner picture is kobuse: a U-shaped high-carbon jacket (kawagane) wrapped around a softer core (shingane). Other constructions—honsanmai, shihozume, and related multi-steel builds—add side panels or spine steels, but the logic stays the same. Put hard steel where the edge must be; put tough steel where a crack would otherwise run through the whole section.

Compare the layers like this. High-carbon jacket steel wants enough carbon to form a dense martensite edge after a fast quench—hard, wear-resistant, capable of a keen apex. Alone, that same steel is crack-hungry under shock. Low-carbon core steel will not hold that razor geometry by itself, but it yields and absorbs energy. During quench, the lower-carbon core is also less eager to transform into a fully brittle martensite network even where cooling is imperfect. The jacket can harden; the core stays the shock path. Modern case-hardening aims at a similar hard-skin / tough-heart idea with different tools. Japanese sword smiths get there with bloom sorting, folding, and forge welding.

Carbon differential is therefore two nested systems. First, composition differential: different billets welded into one cross-section. Second, cooling differential at yaki-ire: clay thickness deciding who cools fast. You can have a beautiful hamon on a simple mono-steel practice blade with clay alone, but traditional art swords usually stack both tricks. That is why museum blades feel “alive” under a loupe: hada from folding, hamon from quench, and a core you only meet if the blade is broken or sectioned for study.

Yaki-ire: clay, heat, and a dangerous dunk

When the blade is shaped—sunobe drawn out, bevels set, curvature planned—the smith coats it with yakiba clay (tsuchioki). Recipes stay workshop secrets, but the ingredients are mundane: fine clay, charcoal powder, stone or iron-oxide powders. Technique is not:

• A thin guide line (hazakai / related boundary coats) sketches where the hamon should run.
• The edge zone gets thin clay or bare steel so water can steal heat immediately.
• The spine and upper body get thick insulating clay so cooling lags.

The blade is heated to the smith's judged austenitizing color—hot enough for the edge carbon to dissolve into austenite—then plunged into the tank in one committed motion (yaki-ire). Edge metal that cools fast transforms toward martensite: hard, expansively stressed, optically “white” after polish. Spine metal that cools slower settles into tougher mixtures often described as pearlite / bainite families rather than a full hard martensite shell. Volume change during the edge transformation also helps pull the famous curve (sori) into a blade that went into the tank straighter.

Nie and nioi—coarse sparkling martensite vs mist-fine brightness—track cooling severity and carbon. Aggressive edge quench leans nie; gentler boundaries deepen nioi. School styles (straight suguha, clove choji, wild midare) are clay handwriting as much as brand logos. A failed quench cracks weeks of work. That is not romance; it is physics meeting humidity, clay thickness, and a human estimate of temperature without a thermocouple glued to the hamon line.

After the tank: why polish finishes the argument

Quench sets microstructure; polish makes it readable and usable. The togishi's stones cut scratches, set the edge geometry, and reveal hamon contrast and hada. Without that work, you have a darkened, stressful bar—not the blade collectors argue about. Forging and polishing are separate professions for a reason: one creates the differential; the other proves it survived.

If you are comparing a museum tachi to a factory stainless wall hanger, you are not comparing “fold counts.” You are comparing whether anyone ever ran a tatara, sorted a kera, welded a soft core, and risked yaki-ire. Our sword-makingguide covers the wider shop flow; this page is the steel's biography from sand to quench.

FAQs