The terminology used to describe historic direct iron smelting furnaces is not uniform, although some influential authors (primarily Pleiner) have attempted to establish a descriptive system. Mostly these are based on slag remains and slag management systems. (Understandably, as it is glassy slag which is often all that remains to examine through archaeology.) In this summary the stress is in the construction of the upper portions of the furnace. The furnace used by the Norse at Vinland, circa 1000 AD, is being presumed to have been constructed with no large tap arch, and to have used the top extraction method.
There are four basic possibilities for how the iron smelting furnace at 'Leif's Houses' (L'Anse aux Meadows) might have been constructed:
1) A clay cylinder that is free standing
2) A (thin?) clay cylinder inside a ring of supporting stones
3) A stone slab box with clay only sealing the corners
4) A series of smaller stones mortared together with loose clay into a cylinder
The Dark Ages Re-creation Company team has built and effectively operated furnaces using all of the first three possible construction methods. In this report, I will offer up samples of historic furnaces constructed using the various methods. (Those images are from Pleiner's 'Iron in Archaeology' or were kindly provided to me by Dr Birgitta Wallace.) One important fact - all the historic examples appear to be from 'industrial' level iron processing sites. Those furnaces are larger (some as large as 60 cm ID), and intended to produce larger blooms per firing than our normal test furnaces. They are also very heavily constructed, clearly intended for multiple firing sequences. Both of these features are to be expected in a location specifically built for full scale iron production. One important consideration is to remember the furnace built at Vinland was only intended for a single use, and is made using a lighter construction to reflect this.
LAM 2 - October 2009. Interior of furnace (tuyere at middle left) near the end of pre-heat.
Lighter coloured flaws on the inner surface are marks where damp clay has spalled off during heating.
The clay debris recovered at LAM do do not indicate any organic materials have been added to the clay used in construction. Our own experience is that without the inclusion of straw (cobb) any clay furnace is certain to suffer extensive cracking. This is primarily due to water within the thick material expanding as it flashes to steam as the walls are heated. Careful drying and slow heating can reduce this effect. The truth is that even still, with any clay thickness over about 2 - 3 cm serious cracking (if not explosive spalling!) is certain to occur.
At LAM, there had been sand added to the raw clay as a 'temper'. Our own experience has shown that the function of such sand addition serves to stiffen up the clay as it is mixed to produce the desired working consistency (balancing water content). Once built and during firing, the sand has not shown to significantly effect the durability or operating characteristics of the furnace walls.
1) Freestanding Clay
Furnace at Lodenice, Bohemia ('Late Romano-Barbarian')
Note that this furnace has a large tap arch, and also is partially earth banked
Although most of our 'Norse Short Shaft' furnaces are built using clay cobb (chopped straw added) there have also been a number of furnaces that were constructed of straight clay. The furnace used for LAM 1 (May 30, 2009) was constructed of freshly mixed potters clay, without any sand added. The walls were roughly 5 cm thick. After a relatively long and gentle pre heat (about 3 hours) there was found to be considerable and heavy cracking to the structure, in two forms. First, as the inner surface close to the fire dried faster and shrunk, it pulled away from the outer layer of clay. The air gap thus produced further reduced the penetration of heat from the interior to the outer layer.
LAM 1 - straight clay, free standing.
Top of the furnace wall, near the end of pre-heating / drying.
The effect of the inner layer drying faster and shrinking is clear to see.
The second effect was long cracks, where major sections of the wall started to separate totally from each other. In most cases, these cracks ran straight through into the interior. The results were extensive enough that several loops of wire were applied to the exterior of the furnace to hold it together. Despite repeated applications of semi-liquid clay slip to attempt to seal these cracks as they developed, they remained in place over the entire smelting process. One actual advantage of these cracks is that they allowed the furnace to self tap excess slag in the later stages of the smelting process. (What Arne Espelund calls an 'incontinent' furnace.) This effect is likely to result in distinctive tap slags, tending to long thin runnels, irregularly spaced around the furnace. (This in comparison to the larger plate shapes seen from the use of slag tapping through a pre-constructed arch.)
LAM 1 - straight clay, free standing.
Late in the smelt, showing larger cracks extending to the furnace base, with hot slag self tapping.
2) Clay with Stone Support
Furnace at Verhurhaugen, Norway.
A relatively thin walled clay cylinder with stone supporting.
The clay cylinder supported by earth or stones with packing has been the primary type developed for past experiments, the 'Boxed Short Shaft'. This basic construction is seen at a number of documented Viking Age iron production sites in Norway .The main advantage of clay cylinder inside stones is that the stones provide support for the clay structure. By packing the gaps between the stones and the clay with earth or sand (or sand / ash mix as we do) the effects of large cracks is minimized. One added advantage to the packing material is that it also effectively will prevent loss of hot gasses, the working chemistry of the reduction process, from escaping.
LAM 2 - October 2009. Furnace about half way through the smelt sequence.
Volumes of gas can be seen venting out of the large cracks due to pure clay construction, despite supporting stones and layer of sand / ash packing.
The clay cylinder that forms the core of the furnace can be significantly reduced in thickness, as the surrounding stones and packing material physically support the structure. The clay walls thus are serving as a containing refractory layer only, so only need to be thick enough to withstand the erosion effects of the high smelting temperatures. Past tests have shown straight clay walls as thin as 5 cm can easily withstand a single smelt sequence.
It should be noted that most of the stones used for this type of supporting construction are very unlikely to show anything but the most minimal heat effects. In our own experiments, it would be difficult to distinguish even the most heavily heat effected from those stones used to ring a simple wood cooking fire.
3) Stone Slab
Furnace from Skeie, Norway.
Showing rectangular shape in stone slab construction
The first of the DARC experimental series used stone slab construction to create a rectangular furnace. The type was returned to for a couple of smelts in 2008. The main reason this construction has had limited use is that suitable stone is not readily available near Wareham. A second reason is that it was seen that even under a forced air blast (electric blower) the burning zone inside the furnace is a D shape, with the flat of the D centred on the tuyere point. This distribution leaves the two rear-most corners of the furnace cold and unignited. The net effect was found to be a reduction of the overall performance of the furnace. Any ore accumulating in those corners is not correctly involved in the reduction chemistry. This results in a significant drop in overall yield and effective wasting of resources.
One further problem with construction using stone slabs is being able to effectively seal the gaps between individual stones. Although clay may be used to attempt this, in practical experience it has been found that as the clay dries, and certainly with the heat of a smelt, the clay shrinks or cracks away from these seams. In an earth banked construction, the surrounding earth limits the loss of reduction gases. In a free standing structure, 'plugging the gaps' becomes a major problem. With a massive furnace, or with skilled hands doing the building, these losses can be greatly reduced or eliminated.
October 2008 - stone slab construction, into the later part of the smelt.
Note the large leakage of reaction gases at the edges of the front stone slab.
Any direct contact of the stones with the interior of the furnace results in an often heavily eroded and certainly distinctively melted inner surface. This effect has been most pronounced in the surfaces closest to the tuyere point. Slag also attaches, often quite heavily, to those surfaces. These are distinctive features, usually over the entire inner surface nearest to the developing slag bowl. Erosion has been seen to a depth of as much as 2 cm (though this is certainly dependent on the type of stone used).
October 2008 - inside surface of the stone slab placed just above the tuyere, seen in the smelt above.
At a point roughly 5 cm above the tuyere, the initial thickness of 2.5 cm had been erroded down to only .5 cm.
4) Mortared Stone
Two furnaces showing 'stones in clay' construction (Sweden?)
Admittedly, we have not attempted to construct a furnace out of stones, using these as if they were bricks, and sealing the gaps with clay mortar. Unless a location was seriously lacking in clay, I can see many more technical problems with this method than any benefit, other than speed of building. First, the clay serving as mortar is certainly going to shrink and crack as it dries, creating serious gaps in the entire surface. These may not present structural problems, but they are certain to void reactive gases from the interior, perhaps to such an extent to 'crash' the developing chemistry. A critical factor in the effectiveness of this method lies with the stones themselves. If a location lacked stone as slabs, but hand suitable igneous stones as rounded pieces, perhaps this method might be employed. Depending on the size of the stones, and the thickness of the desired walls, the clay required might actually prove to be more than that required for solid clay construction. In terms of the physical remains, it would be expected to see much the same effects of high temperatures (melting and erosion) plus slag adhesion seen in straight slab construction.
The primary reason we have not yet attempted this building method is that here at Wareham we lack suitable stone. Our base here is a kind of horrible soft yellow tumbled limestone. Not only will this material never survive smelting temperatures without completely disintegrating, it is so porous I would be afraid we would experience explosive steam effects on heating it.
For a number of reasons, my opinion is that the smelter constructed at L'Anse aux Meadows by the Norse is most likely to be of the 'Clay Cylinder with Stone Support' construction.For that reason, the current Vinland experimental series has concentrated on that type. An initial discussion of why specific furnace sizes and construction methods have been chosen can be found :
A Furnace for Vinland