By: Roger Smith

I must warn the reader that in this first section, I will simply describe the smelting process. A discussion of our ancient and comparative evidence follows.

A bloomery is a furnace in which iron ore is smelted and from which metallic iron is produced. There are a set of complex chemical processes that take place within the furnace which separate the metal from the other materials in the ore. I will omit a technical explanation of the chemical processes, but attempt to describe in detail what a bloomery is and what goes on within the furnace. What follows is a summation of the smelting process, so please keep in mind that I will not refer to specific evidence for processes or raw materials until later.

The base of a bloomery found at the Bardown site.

When the ore has first been removed from the mine, it may be very wet. This moisture would undermine the smelting process, and so the ore must be dried before it is smelted. In order to do this, we can roast the ore just as we would roast a chicken, by subjecting it to heat. The intense heat will drive off any moisture and will also open up the tiny fissures within the ore, thereby increasing the surface area, which will be useful in the smelting process itself. An interesting point that Cleere makes is that the heat required during roasting does not have to be great. He says that between 500° and 600° Celsius is sufficient. This temperature is easily obtainable with the proper fuel and ventilation.

A roasting pit.

Cleere carried out the roasting process with a reconstructed roasting bed (based on evidence from the Bardown region). Cleere found that by alternating layers of ore and charcoal, he could dry the ore very efficiently. With the ore and the charcoal spread out evenly, Cleere ignited the whole mass at one end and allowed it to burn through to the other end. In Cleere’s experiment, the roasting process took about a night. However, once the ore was roasted and allowed to cool, it was sieved to remove the smaller pieces of ore. This was done because, as Cleere explains, if these smaller pieces of ore were allowed into the smelting process, they would clog up the ventilation holes. For this reason, the smaller pieces were removed; we know this because "deep beds of roasted ore particles below 5 mm are common finds on refuse heaps of early Wealden sites."(Cleere, pg.35) We have just expelled the moisture form the ore and can now begin the smelting process.

A crude bloomery.

The ore that comes from the mine consists of iron carbonate. To move from the iron carbonate and other impurities to pure iron, we must separate the iron from the carbon and oxygen. However, iron and oxygen have a very strong affinity for each other. Therefore, we must provide the oxygen with another molecule that it likes more. CO(carbon monoxide) is that molecule and (The CO comes from the combustion of peat or wood ro charcoal) during the roasting process, heat is applied to the FeCO3, yielding FeO and CO2. The CO2 goes off in the form of a gas, while the FeO attracts more oxygen to form Fe2O3. This is what is used in the bloomery.

So, we begin the smelting process with Fe2O3. To remove the oxygen from the Fe2O3 is very difficult because iron oxides are very stable compounds and cannot be reduced without extreme heat; heat that would have been impossible for the Romans to obtain with the fuels they used. Instead of employing heat alone, however; we may add a reducing agent to attract the oxygen away and leave pure iron. "The reducing agent used in the bloomery process is carbon monoxide (CO), resulting from the combustion of carbon."(Cleere pg. 32) The carbon monoxide combines with the oxygen from the ferric oxide(Fe2O3) to form carbon dioxide (CO2) which goes off as a gas, leaving pure metallic iron (Fe) behind.

Now that we know the chemical processes that take place within the bloomery, we can focus on the physical aspects of the bloomery

The primary fuel in ancient bloomeries was charcoal (burnt wood or reeds). Peat and coal could have been used and were occasionally, but they did not burn hot enough, according to Cleere, and so charcoal made from a hardwood, such as oak or ash, was used because it was free from impurities and could produce the temperature required to smelt the iron. With the proper fuel chosen, we must now move to the construction of our bloomery. "The furnace needs to be airtight, so as to prevent the access of undesirable air into the interior."(Cleere 1976, 132) If oxygen is allowed to enter the furnace it will recombine with the reduced iron(Fe) and we will end up with iron oxide again. According to Cleere, the best material to keep oxygen out of the furnace is clay. So now we have the fuel(charcoal) and out airtight bloomery, and all we need is to fire it up.

Cleere, having reconstructed a Roman furnace, preheated his experimental furnace for half a day and sealed up the front opening of the bloomery with a wedge of turf, sticking a tuyere in the bottom of the furnace. Charcoal would be fed in through the top of the furnace and bellows would be inserted into the tuyeres to bring the temperature inside the bloomery up to around 1300° Celsius. (You may be asking yourself how charcoal can be inserted into the bloomery through the top, if the bloomery is sealed airtight. Well, the bloomery takes advantage of the fact that heat rises, so if the bottom of the bloomery is sealed tight to prevent oxygen from coming in, then as the heat rises through the top of the furnace, no oxygen from outside will be able to enter, thereby allowing the reducing process to take place). "Once this temperature has been obtained, small additions of ore and charcoal, probably no more than 2 lb. at a time, would be put into the top of the shaft."(Cleere 1976, 134) The ore moves slowly down the furnace shaft, through increasing temperatures, until it collects in a spongy white-hot mass at the bottom of the furnace. The burning charcoal and the oxygen supplied by the bellows produces the reducing agent, carbon monoxide(CO), which reacts with the ore when the two come into contact, drawing oxygen atoms from the iron oxide and forming carbon dioxide(CO2) which leaves as a gas, while free iron particles remain and collect at the base of the furnace along with some impurities of silica(also known as gangue) to form a slag.

"Meanwhile, the iron oxide began to react with silica from the ore to form a slag in which the many iron particles were trapped. The formation of the slag protected the iron from reoxidizing as they passed in front of the tuyere, where the atmosphere was extremely oxidizing. Below the tuyere, some slag drained out of the pasty mass of slag and iron particles and fell to the furnace hearth. The final bloom was a porous lump of iron, somewhat refractory silicates and bits of unreduced ore that collected above the liquid slag. Overall it was heterogeneous in carbon content, and it was, generally speaking, steel."(Cleere 1995, 526)

The bloom could be removed through the front archway or through the top of the bloomery. The amount of iron would depend upon the time spent and the amount of ore added to the bloomery. Cleere produced about 10 kg of iron in one day, and he concluded that a skilled Roman smelter could produce at least twice this in one day.

The smelting process.

Once the bloom was removed from the furnace, Cleere theorizes that the bloomery would be relined, patched, and immediately fired up again so as to save the heat that is in the walls of the bloomery.(Cleere 1976, 136) The bloom(called an unworked bloom) would have pieces of slag within it that would have to be removed by a smith. To do this, the smith would heat the bloom to about 1100° Celsius and would hammer it repeatedly to pound out the molten slag(At this temperature, the slag would be molten because it has a lower melting point than iron.) The end product is called a worked bloom.

With the properties and processes of a bloomery defined, we can now move on to consider how we know all this. I will begin with the ancient evidence.

Ancient Evidence

Cleere finds most of his ancient evidence in the Weald, a district in the middle eastern part of modern England where at least 67 ancient iron-smelting sites have been identified. Of all those sites, I will focus on two: Bardown, located in mid-eastern Britain, and Holbeanwood, located very close to Bardown.

We know the Romans used roasting pits because they are sometimes found at iron-working sites, specifically in the Weald. As Cleere tells us, "the Bardown settlement provided two examples of a specially constructed hearth that was indisputably used for ore roasting."(Cleere. pg35) We think they were used for roasting because they have patches of burnt soil, along with scraps of ore, surrounding the pit. Cleere gives us the dimensions of these pits at Bardown. He says they were typically about 2.5 m long by 0.8 m wide and 0.2 m deep, surrounded on the sides with stones covered in puddled clay. Cleere also mentions that "refuse heaps" containing small pieces of ore less than 5 cubic mm are common finds at these sites, and that too gives evidence concerning the process of preparing ore for smelting. Cleere infers that these small particles of ore were excluded from the bloomery process because they would clog up the tuyeres. Thus, we have figured out how they roasted ore, but what did they use for fuel?

As would be expected, there is little evidence for the type of fuels used: they were, after all, burned up in the process. Cleere does mention that classical authors indicate that the charcoal used in ancient times is of the same type that is still in use today. In addition to the literary evidence, Cleere found small pieces of charcoal near the roasting hearths, and it seems safe to assume that charcoal was the fuel used in both roasting and smelting which gives us more insight into the fuel used. But what did the bloomeries look like?

No one has found a perfectly intact Roman bloomery yet, but we can learn a great deal from the bases of the bloomeries which still remain. The type of furnace I will focus on is the Roman shaft furnace. There are other types but this seems to be the most commonly used furnace in Roman times. Cleere has found examples of these in Holbeanwood and Minepit Wood. "They consist essentially of clay shafts, about 0.3 m in internal diameter with walls 0.25 to 0.3 m thick at the base. Their original height can be inferred to at least 1m."(Cleere pg. 41) When Cleere reconstructed a furnace using one from Holbeanwood as an example, he found that if the front archway was sealed with clay, when it was time for the slag to be removed, the seal would have to be broken, hence reducing the temperature within the bloomery. So Cleere used a piece of turf to block the front archway. When it was time for the molten slag to be tapped, the turf would be removed and the molten slag would run out. However, this leaves two problems: 1.This doesn’t give us any insight as to how the Romans made the front archway. 2.Won’t the temperature drop just as much if you use a piece of turf as it will if you use clay?

I can probably answer both questions at the same time. Cleere found post holes on one side of the furnaces at Holbeanwood and he suggests that they were used to hold the bellows above the molten slag as it was pouring out so that hot air could still be moving through the furnace while the slag was draining. We may guess, then, that the Romans used a temporary archway consisting of a piece of turf (or something like it).

While intact bloomeries are few and far between, tuyeres are a plentiful find on Roman smelting sites, especially at Bardown. It is pretty clear from the slag on the tips of the tuyeres that they were inserted far into the furnace, in order to raise the temperature to very high levels. This is merely inference, but we have comparative evidence, in the way the Hayan tribe uses the tuyeres in Africa, which supports our conjecture.

Comparative Evidence

Comparative evidence abounds just as much as ancient evidence. To begin we will first look at the Hayan people in Africa, where the process of smelting is still known by the elder tribesmen, although it is no longer used much. From the Hayans, who have archaeological evidence of iron-smelting that dates back to 200-600BC, we are able to supply some of the pieces missing at the Roman sites and assume that the Romans did it in the same manner.

According to Peter Schmidt, who observed as the Hayans carried out the smelting process, the Hayans felled trees which they burned to produce charcoal used to roast the ore. The roasting took place a day before the smelting and was done to remove the moisture from the ore and to open cracks and fissures in the ore. While the ore was roasting, the tribesmen built the smelting furnace. Schmidt tells us that the "smelting furnace was lined with mud from the earth of a termite mound."(Schmidt pg. 527) The pit was filled with swamp grass that was burned in the pit to produce the charcoal necessary for the reduction process. "Eight tuyeres 50 or 60 cm long were placed around the bowl, with their ends extending deep into it."(Schmidt pg. 527) The upper cone section of the furnace was constructed of slag left over from other smeltings and from termite earth. Once the shaft was completed, a mixture of roasted ore and charcoal was poured slowly in through the top of the furnace. Schmidt notes three characteristics of the Hayan process: 1. "the ore was roasted before it was smelted", 2. "the furnace bowl was filled with charred swamp reeds", and 3. "the tuyeres extended deep into the bowl"(Schmidt pg. 527) (These are all characteristics of the Hayan smelting process that Schmidt finds noteworthy). The final product of the Hayan process is quite distinct in chemical composition from the iron of the Roman, due to the charcoal used and the ore used, but other than this the processes seems to be identical.

Filling In The Blanks

Now that we know the ancient evidence and the comparative evidence, how do they relate?

One of the questions we as archeologists must answer is what were the clay nozzles(tuyeres) found on the Roman smelting sites used for? The Hayans used the tuyeres to protect the bellows from being burnt when they were inserted into the furnace, and that is exactly what Cleere concluded about the Roman tuyeres. Further, the Hayans inserted the tuyeres deep into the furnace pit, and that agrees with the ancient evidence where we find slag vitrified on the end of tuyeres. So the Romans probably also inserted the tuyeres deep into the bloomery. (We can never be absolutely sure when looking at incomplete evidence). But in order to use the tuyeres, there must have existed bellows of some kind. There is no ancient evidence at ironmaking sites to suggest the use of bellows. (This makes sense, because they were probably made of wood and decayed over the years). We can see from the Hayans that bellows are used and are necessary to make the furnace hot enough to reduce the iron. We know that bellows did exist in the ancient Roman world from depictions on stone, so it is safe to conjecture that bellows were used in smelting.

As far as roasting, we find sites of burnt patches of ore and small pieces of coal in the ancient world. And we see that the Hayans roast the ore before it is smelted, to remove moisture and open up cracks. From the mass of ore and coal around the burnt patches and from comparative evidence, Cleere concludes that these burnt patches were used for roasting. A fully intact bloomery has not been found on these ironmaking sites yet, and so the tops of these furnaces are reconstructed merely on the basis of inference and from what we see the Hayans doing today. The Hayans build up the furnace in a conical shape, and Cleere has argued that all effective Roman bloomeries consisted not just of a bowl, but of a domed or cylindrical shaft superstructure(Cleere 1976, 132-134).

Here ends the section on bloomeries.


1.Cleere, Henry and david Crossley. The Iron Industry of the Weald.
Leicester University Press, 1985.
2.Schmidt R. Peter and S. Terry Childs. "Ancient African Iron Production."
American Scientist 83:524-533.
  3.Cleere, Henry. "Ironmaking." Roman Crafts. Ed. Donald Strong and David
Brown: Gerald Duckworth and Co., 1976.127-140.
Please email me:

 Visit UNC-CH