This is a single page in a larger project,
Introduction to the archaeometallurgy of iron
Donald B. Wagner

Co-fusion steelmaking

Joseph Needham coined the term ‘co-fusion’ for steelmaking methods which involve mixing cast iron with wrought iron. The term has attained general acceptance. The apparently simple idea of mixing irons of different carbon contents to obtain an intermediate carbon content brings with it a variety of technical complications, for which steelmakers have used different approaches.

The earliest description of a co-fusion steelmaking process occurs in the biography of Qiwu Huaiwen 綦母怀文, who was active in the +6th century:

Qiwu Huaiwen, whose place of origin is unknown, used Taoist techniques in the service of Shenwu-di 神武帝 [Gao Huan 高欢, 496–547, father of the first Emperor of Northern Qi 北齐].  . . .

Huaiwen made sabres [dao 刀] of ‘overnight iron’ [su tie 宿铁]. His method was to anneal [shao 烧] powdered cast iron [sheng tie jing 生铁精] with layers of soft [iron] blanks [ding 铤, presumably thin plates]. After several days the result is steel [gang 刚 = 钢]. Soft iron was used for the spine of the sabre. He washed it in the urine of the Five Sacrificial Animals and quench-hardened it in the fat of the Five Sacrificial Animals. [Such a sabre] could penetrate thirty armour lamellae [zha 扎] [i.e. thirty layers of armour?]

The ‘overnight soft blanks’ [su rou ting 宿柔铤] cast today [in the Sui period?] by the metallurgists of Xiangguo 襄国 represent a vestige of [Qiwu Huaiwen’s] technique. The sabres which they make are still extremely sharp, but they cannot penetrate thirty lamellae.


怀文造宿铁刀,其法,烧生铁精以重柔铤,数宿则成刚。以柔铁为刀脊,浴以五牲之溺,淬以五牲之脂,斩甲过三十札。今襄国冶家所铸宿柔铤,是其遗法,作刀犹甚快利,但不能顿截三十札也。怀文又云:「广平郡南干子城,是干将铸剑处,其土可莹刀。」 (北史,卷 89).

As usual we should like to know more about the background of this source, but, again as usual, no more information seems to be obtainable. Perhaps the best guess would be that the original source for the description came from Xiangguo, a place near modern Handan, Hebei. The name was used for a district (xian 县) and a commandery (jun 郡) at some times in the Sui period and before, but not later. This was a major iron-producing region in both the Tang and the Song.

Shen Gua 沈括 (1031–1095) describes two steelmaking methods in his Mengxi bitan 梦溪笔谈. His account starts with a description of the making of what he calls ‘false steel’, which is obviously a co-fusion process similar to Qiwu Huaiwen’s:

What the general run of blacksmiths call ‘steel’ [gang tie 钢铁] [is made as follows]. Take wrought iron [rou tie 柔[=鍒]铁], bend and coil it, and insert cast iron into the interstices. Seal with clay, ‘refine’ [lian 炼] it, and hammer to cause [the soft iron and the cast iron] to interpenetrate. [The product] is called ‘combination steel’ or ‘irrigated steel’.

世间锻铁所谓钢铁者,用柔铁屈盘之,乃以生铁陷其间,泥封炼之,锻令相入,谓之“团钢”,亦谓之“灌钢”。此乃伪钢耳,暂假生铁以为坚,二三炼则生铁自熟,仍是柔铁。然而天下莫以为非者,盖未识真钢耳。余出使,至磁州锻坊,观炼铁,方识真钢。凡铁之有钢者,如面中有筋,濯尽柔面,则面筋乃见。炼钢亦然,但取精铁,锻之百余火,每锻称之,一锻一轻,至累锻而斤两不减,则纯钢也,虽百炼不耗矣。此乃铁之精纯者,其色清明,磨莹之,则黯黯然青且黑,与常铁迥异。亦有炼之至尽而全无钢者,皆系地之所产。 (梦溪笔谈, 卷 3).

Still another description of a co-fusion process is by Song Yingxing 宋应星 (17th century) in his Tiangong kaiwu 天工开物:

The method of ‘refining’ steel [gang tie 钢铁]: Wrought iron is beaten into thin strips as broad as a finger and about 1½ cun 寸 [5 cm] long. The wrought iron strips are tightly tied together in a bundle, and cast iron is placed on this. (In Guangnan 广南 [south China] there is [a type of cast iron] called ‘lump raw steel’ [duo zi sheng gang 堕子生钢] which is especially suitable.) The top [of the bundle] is covered with old straw sandals (plastered with clay so that they do not quickly decompose), and the bottom is plastered with clay. [This package] is lowered into a fiery furnace, and the strength of the fire is blown with the bellows. After a certain interval the raw steel first melts and then soaks into the wrought iron, and the two natures [qing 情, of the cast and wrought iron] commingle. It is then removed and hammered. It must be ‘refined’ many times over; once is not enough.

This [steel] is popularly called tuan gang 团钢, but it is more correct to call it guan gang 灌钢.

凡钢铁炼法,用熟铁打成薄片如指头阔,长寸半许,以铁片束包尖紧,生铁安置其上,(广南生铁名堕子,生钢者妙甚。)又用破草履盖其上,(粘带泥土者,故不速化。)泥涂其底下。洪炉鼓鞴,火力到时,生钢先化,渗淋熟铁之中,两情投合,取出加锤。再炼再锤,不一而足。俗名团钢,亦曰灌钢者是也。(天工开物 • 五金 • 铁)

Shen Gua and Song Yingxing use the word lian 炼、鍊, which has both a general meaning, ‘to purify’, and a large number of very specific technical meanings in particular contexts. It therefore corresponds very well to the equally vague and irritating English word ‘refine’, and this is the standard translation for it. In this case lian corresponds to shao 烧 in the first text. The basic meaning of shao is ‘to roast’, and here it clearly means ‘to anneal’, i.e. to subject a metal object to a high heat for an extended period.

Each of these texts describes a process in which wrought iron and cast iron are heated together. In the first the process takes several days, while the second gives no information on the time required. What happens in these processes depends on whether the annealing temperature is above or below the eutectic temperature of the iron, ca. 1147°C.

Lower-temperature co-fusion

Below the eutectic temperature of the iron, the process is rather like the familiar cementation process, with cast iron as the carbon donor: the basic reactions are C + CO2 → 2CO at the surface of the cast iron and 2CO → C + CO2 at the surface of the wrought iron. Solid-state diffusion in both materials carries carbon from the interior of the cast iron and to the interior of the wrought iron. Since this is a diffusion-limited process, the time required is determined by the thickness of the cast and wrought iron and by the temperature. Qiwu Huaiwen’s process is said to require an anneal of several days, and we can therefore assume that it used a fairly low temperature.


Micrographs from P. Whitaker’s experments

Reproduced from Needham 1958, Plate 28.

A possible reconstruction of Qiwu Huaiwen’s method was the subject of experiments in 1955 by Mr P. Whitaker, on the initiative of Joseph Needham. Strips of wrought iron, 1 mm thick, were piled up and wired together with varying amounts of crushed white cast iron between layers. This faggot was heated in a non-oxidising atmosphere to a temperature between 950° and 1000° C, then immediately withdrawn from the furnace and hand-forged, welding it into a single bar. At the original welded surfaces in this bar there were highly carburised zones as well as some unchanged white cast iron. A homogenising anneal of 8 hours at 900° C resulted in a uniform carbon content of 0.8% throughout the bar.

This proposal fits reasonably well with the little we know of Qiwu Huaiwen’s method, though the original description includes no mention of forging the faggot into a bar at the beginning of the anneal. The forging undoubtedly speeds up the process considerably, but is unlikely to be a strict necessity. Another aspect is that the wrought iron strips were very thin in these experiments. An ancient smith would certainly have been capable of producing wrought-iron strips which were only 1 mm thick, but this would have been time-consuming, and would have involved the loss of a considerable amount of iron through oxidation. The ‘wrought iron blanks’ of the description would no doubt have been thicker, perhaps 2–3 mm, necessitating a longer annealing time.

Higher-temperature co-fusion

If the temperature is above the eutectic, the wrought iron is annealed in a bath of liquid cast iron. See the iron–carbon equilibrium diagram, here on the right. Two processes occur, the balance between them depending on the exact temperature and carbon content. If for example the temperature is 1250°C and the carbon content of the cast iron is 4.0%, wrought iron (0.1–0.3% carbon) will be dissolved until the bath is diluted to a carbon content of ca. 3.5%. After this, carbon diffuses into the wrought iron; still at 1250°C, an equilibrium is quickly reached in which the carbon content at the surface of the solid iron is ca. 1.5% and the carbon content of the liquid iron is ca. 3.5%. As carbon diffuses into the interior of the solid iron, using up the carbon in the liquid iron, more solid iron (with 1.5% carbon) precipitates from the bath until no liquid is left. A few minutes at this temperature would result in an inhomogeneous lump of steel with carbon content varying between ca. 0 and 1.5%. Perhaps an hour at the same temperature (the time would depend on many unknown parameters), or many hours at a lower temperature, would homogenise the lump: if the original charge had been 20% cast iron and 80% wrought iron, the result would be a steel with 0.8% carbon, excellent for many purposes.

It is not impossible that Qiwu Huaiwen used a higher-temperature co-fusion process like the one I have described here, though a lower-temperature process seems more likely in his case.

What Shen Gua describes, in the second quotation above, might well have been a higher-temperature co-fusion process, and the intended result might in fact have been a non-homogeneous steel which could be used in making pattern-welded swords. The ‘bending and coiling’ of the wrought iron seems otherwise superfluous. Elsewhere Shen Gua refers to steel swords with the sort of pattern that might result:

Among the names of ancient swords, . . . Yuchang 鱼肠 [‘Fish Gut’] was what today is called a pan gang 蟠钢 [‘coiled steel’] or song wen 松文 [‘fir patterned’] sword. If one takes fish, bakes them, and strips off the ribs to reveal the guts, they have a distinct resemblance to the pattern on a modern pan gang sword.

古剑有沈卢、鱼肠之名,沈音湛。沈卢谓其湛湛然黑色也。古人以剂钢为刃,柔铁不茎榦;不尔则多断折。剑之钢者,刃多毁缺,巨阙是也。故不可纯用剂钢。鱼肠即今蟠钢剑也,又谓之松文。取诸鱼燔熟,褫去胁,视见其肠,正如今之蟠钢剑文也。(梦溪笔谈, 卷 19)


The possibilities of ‘pattern welding’ steel and soft iron to make intricate patterns have been exploited by swordsmiths in Europe and Japan as well as in China. The pattern is made by variations in carbon and phosphorus content, which become visible when the sword is polished and etched. He Tangkun (2001) has examined several patterned-steel swords in private Chinese collections, and gives the three photographs shown here.

What sort of furnaces were used in these processes, and how was the metal held together while it was heated? Were crucibles used? The experiments by P. Whitaker, noted above, showed that a lower-temperature co-fusion process can be carried out without a crucible or other container for the material. In those experiments a modern electric furnace was used, but the same conditions – a temperature in the range 950–1000°C in a non-oxidising atmosphere – is regularly attained and held for long periods in a charcoal-fuelled smithy hearth. A furnace specifically designed for heating crucibles would be easier to control and much more fuel- and labour-efficient, but would not, strictly speaking, be necessary. Such a furnace was used in co-fusion steelmaking in Persia in the 19th century, and was described in detail by a Russian, Captain Massalski.

In Shen Gua’s description of a higher-temperature co-fusion technique he states that the mixture of cast and wrought iron is sealed with clay. Perhaps he meant that the material is placed in a crucible whose mouth is then sealed. On the other hand an equally plausible interpretation is that the material is plastered all over with wet clay to hold it together and protect it from the furnace atmosphere while it is heated to the necessary high temperature. When the co-fusion is complete and hammering begins, the hard-fired clay breaks up and the steel remains. The heating can easily be done in a smithy hearth.


He Tangkun 何堂坤. 2001. ‘Ji mei chuanshi huawen gang jian de chubu kaocha 几枚传世花纹钢剑的初步考察’ (Preliminary investigation of several patterned-steel swords handed down from ancient times). Wenwu 文物 2001.1: 68–73.

Needham, Joseph. 1958. The development of iron and steel technology in China. London: The Newcomen Society. Second Dickinson Memorial Lecture to the Newcomen Society, 1956.

Last edited by DBW 24 February 2023