LAM, Lower & Longer : June 25, 2022

Reduced air Volume
'Vinland' clay test

Introduction :

For the first smelt of 2022, a new furnace was constructed, a slightly smaller diameter 'Norse Short Shaft' set on a plinth. The smelt was conducted applying an air volume based on the measurements from the October 2021 'Wind' testing. The secondary experiment was the inclusion of a patch of furnace wall made up of materials gathered at L'Anse aux Meadows, Newfoundland (close to the Norse 'Vinland' archaeological site).

Part One - The Smelt :

The furnace constructed for these experiments was the long proven ’Short Shaft’ type. (1) In this case the furnace structure was placed on a rectangular plinth formed from 'half thick' concrete blocks, centre filled with charcoal fines. The blocks stood 18 cm tall and created an internal square at roughly 26 cm.
The majority of the clay walls were constructed of EPK powdered clay, course sand, and dry shredded horse manure (rough thirds by volume) again a standard mixture. Individual fist sized ‘bricks’ were placed against an internal metal form, resulting in a 25 cm internal diameter. The walls tapered from 6-7 cm at the base to 4 cm at the top. During the build, the exterior was wrapped in rope to reduce sagging. To both support the structure and promote drying, the interior was filled with a mixture of sand and wood ash.

Limestone blocks laid out for the plinth Completed build, rope re-enforcing
The 'Vinland' clay is the darker material

The initial construction would result in walls built to an average of 60 cm height (varied 59 - 62 cm). The scheduling was collapsed, with the build undertaken over the afternoon the day before the smelt, and initial drying fire that same evening. This would lead to more than typical cracking in the furnace wall, primarily running along the joints between the individually applied 'bricks'.

drying crack crack
During drying fire - tuyere set After drying - major crack above extraction arch Vertical crack does not extend through interior

Because of the number of smaller cracks (most not extending into the interior) it was decided to place a series of loops of soft iron wire around the body of the furnace, placed about every 10 cm.

The tuyere was the heavy forged copper proven over many previous smelts. It was set as has been shown effective :

25 degree down angle
5 cm proud of interior wall
16 cm from 'soft' base to centre
44 cm stack height

Some of the charcoal fines had been consumed during the drying process. A layer of ash/sand mix at 3 cm thick was applied to return the base level to the top of the concrete blocks. A typical extraction arch was cut (to 10 cm tall by 18 cm wide).

rough scaled
Rough field drawing by Rey Cogswell (with additions) Converted scale drawing

Part Two A - 'Vinland' Clay

During a trip to L'Anse aux Meadows NHSC, permission was given to gather a small amount of clay from an existing bank at Norstead, situated about 800 m NE of the archaeological site. Further, some sand was collected off the beach of Medee Bay, roughly where the small brook that cuts the clay bank empties into the ocean.
This represents the natural clay located closest to the site of the circa 1000 CE iron smelting furnace built by the Norse, as well as the basalt based sand locally available. (It should be noted that this material has not been matched via chemical analysis to the furnace wall fragments excavated at the site.)

clay sand
Raw clay from the Norstead bank
Basalt sand from the Norstead beach

A fairly quick look over the known location of the clay bank was made (done in very foul weather, even for 'spring' in North Newfoundland). No special preparation of the clay material was undertaken, other than pulling out any roots or larger stones that were found during blending. The sand included many small white pieces, broken shell fragments. No additional water was added for mixing, so the sand at least was damp with sea water. A quantity composed of 1793 gm clay and 1000 gm sand was hand blended together. The result was about the same consistency of the normal mixed clay material, but was found to be less plastic and more prone to breaking over 'squishing' as manipulated.

A test patch of this material was applied during the build, located roughly 45 degrees to the tuyere, to the opposite side from the extraction arch :

starting 6cm up from base of furnace (so 10 cm below tuyere centre)
positioned back 14 cm (exterior) / 9 cm (interior) from the tuyere
patch was 10 cm wide
patch was 23 cm tall (to 29 cm / 13 above tuyere)

A set of four holes was cut through the furnace wall, bracketing the test patch, of a size that would allow for the insertion of a thermocouple:

two lower holes were 14 cm from the base (so about 2 cm below tuyere centre)
two upper were 30 cm above base (so 14 cm above tuyere centre)
two towards the front were 12 cm from the tuyere (outer measurement)
two towards the rear were 32 cm from the tuyere (outer measurement)

The position of the ports were placed specifically to give an indication of the temperatures the internal surface of the test patch were subjected to. Measurements were made several times over the smelt, using a Omega HH12B digital pyrometer, equipped with a rigid thermocouple, rated for 1330 C.

Using thermocouple, time point 1035

Part Three - the Smelt

burn ore
Air supply, blower with gate to rear Part way through main sequence Ore addition (at later 3kg charge point)

Team :

Darrell Markewitz = (furnace construction) smelt master
Neil Peterson = records, ore measure and compaction
Maxim Fedetsov = charcoal, charging and compaction
Rey Cogswell
= charcoal and extraction
Isabelle Wigglesworth = charcoal and compaction
Richard Schweitzer = compaction

Sequence :

    The sequence data is available here.

Normal practice is to fill first with rough (unsized) charcoal, then once the fire is well established, switch to graded (screened at 5 to 25 mm) fuel. Once the ignited portion has reached the upper part of the furnace, and the exhaust gases naturally touch off, additions will start. Normally a quantity of broken iron rich slag, typically about 3 - 5 kg is added initially, to quickly establish a working slag bowl system to allow the collection of reduced ore as bloom. For this smelt, there was no previously collected iron slag available, so in place was added 2 kg of previously prepared silica rich and iron poor limonite (Barton's Run).
As our ore analog was added, amounts were increased in a uniform sequence, with three additions of each amount : 1 / 1.5 / 2 / 3 kg, distributed evenly through each volume of charcoal.
Slag was tapped three times over the smelt, the first two (at 14:31 and 1526) as indicated by changes in sound of the air blast combined with visual appearance down the tuyere. The third was a considerably larger volume, and was a self tap of the furnace at 1643.

Air :

The standard air supply has been provided by a compression styled high volume electric blower (purchased in 2008 as US Navy surplus). This unit is rated to 1400 LpM, considerably more than required for past smelts in the typical 25 - 30 cm ID furnaces. During the 2021 'Wind & Weathering' smelt, a series of air volume measurements were undertaken, using a more accurate Omega HHF 1000 flow meter, providing continuous output linked to a lap top for recording. Part of that experiment included measuring air delivery using the previously constructed 'smelt bellows', over three different production cycles, each with turns taken by three different operators. (3) Overall, the average produced was 520 LpM (based on a stroke count of 60 per minute).
For that reason, the output of the electric blower was reduced to approximately 500 LpM, as indicated by the scribed lines on the blast gate. These were marked a number of years and many experiments ago, lines which were at best rough indications of the values. Comparison of the marks against actual flow volumes as measured by the accurate meter in 10 / 2021 suggested that the actual delivered air was roughly 85 % less than the plate marks. Unfortunately a less accurate (and somewhat problematic) instrument was brought for this experiment, rather than the new (and accurate) Omega unit.
Taken together, the actual air volume employed is estimated to 425 LpM (very roughly!). Working from the well proven guidelines for air volume against internal area documented by Sauder and Williams, the 'ideal' air volume for a 25 cm ID furnace would be between 590 to 880 LpM.
The primary impact of this reduction in delivered volume was seen in the considerable increase in charcoal consumption rates, at an average of 24 minutes per 'standard bucket' of 1.8 kg. / time per 1 kg at just over 13 minutes.

Temperatures :

Because the expectation was that the interior of the furnace might approach as high as 1350 C, the probe was removed at roughly 1270 + C, while numbers were still increasing. At one point distraction meant the probe was left in place too long, and the stainless steel shell on the probe actually melted off, suggesting the temperature at that point and location had well exceeded the rated temperature of the thermocouple of 1330 C. The first set of measurements were made only 35 minutes after the first addition of charcoal, it is clear this furnace quickly ran up to suitable operating temperatures. The 11:25 (clock) time, about an hour after first charcoal, was just before the first addition of the Barton's Run material. The 12:24 time just before the first addition of ore analog.

TIME (clock)

bottom front
1270 +
1275 +
top front
1275 +
bottom rear
1270 +
1275 +
1270 +
top rear

Table A : Charting the interior temperatures

Extraction :

The furnace had been constructed on a block plinth, specifically to make extraction easier.
The clay door to the extraction arch was pried loose, and the two front blocks swung open. This allowed for the bulk of the lower layer of charcoal fines to be raked clear. Initial probing indicated that the slag bowl had filled the bottom of the furnace, and so was attached completely to the walls. With space below, the log 'thumper' was used to quickly beak the entire slag bowl down and free. Although a quick process, this did result in a considerably larger mass than normal to pull and carry over to the stump for compaction.

Video Sequence of the full consolidation process,
by Anita Herbert (used with permission)

hammer hammer
Starting compaction, slag bowl still attached Part way through compaction, encasing slag obvious

A considerable surprise was the size of the final bloom at 7 kg. The material remained fairly spongy in consistency, likely with some internal slag. The produced yield was 28 %, considerably higher than expected based on past experience.

Final bloom, top surface, likely tuyere to SE side

The image above was taken at the end of the consolidation hammering. (Beware the extra brightness applied by the camera, the actual heat colours were medium orange to dull red) The film of slag still attached are clear as the dark areas over the hotter metal core. The upper surface of the bloom had a distinctive hollow in the centre. This is sure to have been caused from both the slag bowl and the collected bloom sitting too high in the furnace, resulting in the air blast cutting into the top surface and eroding it. This all is likely an effect of the lower than ideal air flow, which did not penetrate as deeply into the bottom of the furnace. The effect might have been reduced if some of the fines had been scraped out, allowing the developing bowl to settle downwards midway through the smelt. (Note that this suggests that with better control, the bloom might actually have been larger!)

Part Two B - 'Vinland' Clay

Did the 'Vinland' Clay survive use as a wall material in a full iron smelting sequence?
The answer is a 'qualified' yes.

During build and drying sequence, the Vinland clay proved a bit difficult to blend uniformly to the standard EPK mixture. Perhaps predictably, the Vinland clay mix was found to contract at a different rate, producing a crack that ran along it's margins. This was repaired simply enough by just adding additional clay into those cracks. There was no actual failure of the Vinland material, in terms of cracks venting furnace gasses or burning through during the smelt.

patch patch
interior, showing entire 'Vinland' clay area detail, lower portion of 'Vinland' clay area

When the interior of the furnace was examined the following day, it was clear that the Vinland clay had suffered significantly more heat erosion effects, basically starting to melt. In the images above, it should be noted that the hottest part of the furnace is located in a rough oval around the tuyere, typically extending 10 cm to either side and bellow, plus to 15 cm above. In the detail image above, there is a clear difference between the appearance of the EPK (virtually no effect) compared to the Vinland clay at the same height. This despite the fact that the Vinland test material was positioned just beyond the normal 'hot spot'.

When the exterior surface of the furnace was examined however, the results of furnace use were different.

Vinland test patch after firing
After handling of surface

At first glance, the Vinland patch looked to be in about the same condition as the rest of the furnace walls. there was a clear deep crack running around the whole boundary of the patch, but this was not considered unexpected, as differential shrinking had already shown these cracks. The surprise was when the test material was examined more closely.

Even a slight pressure (hand rubbing) caused the Vinland material to completely crumble apart to a sand like texture. The portions towards the interior side had fused together, melting into a thin shell roughly 5 mm thick, where not burned away entirely.
It was clear that the area of the test patch would not endure a second firing. Major repairs by adding more of this material are considered questionable, as the loose sandy surface would make any binding of new clay mix to the existing surface difficult (if possible at all).

Part Five - Conclusions :

1) Reduced Air

What the exact difference between air from a human powered, Norse type (double bagged) bellows and an electric blower remains unclear. There are a number of design elements that potentially effect the nature of air delivery from such a bellows, with produced total volume only the most obvious. The first part of this experiment was intended to test just that single aspect, by reducing total available volume down to an amount recorded from a Norse bellows of a size considered physically realistic to operate over the long duration of an effective smelt. A number of experimental smelts were undertaken in the past using similar sized and constructed furnaces, slight variations on the same ore analog, with the same bellows unit used for the initial measurements reported here. (4) The results however were without exception significantly different, with smaller blooms produced and much lower yields (in the range of 3 - 5 kg and 15 - 20 % yields)

This team has long worked within the air volumes suggested (and long illustrated) by Sauder & Williams, given as 1.2 - 1.5 litres per cm 2 furnace internal cross section at tuyere - with good results. (5)

June 2022
October 2021
Ore Amount
24.5 kg
24.3 kg
7 kg
7.6 kg
28 %
31 %
Furnace ID (at tuyere)
Air Volume
425 LpM (estimated)
750 LpM (average)
Air per cm2 (at tuyere)
0.86 LpM
1.22 LpM
burn rate (average)
13 M/kg
8 M/kg

Table B : Comparing experiment 91 to 90

The full impact of the reduction in air volume was likely somewhat obscured by the decision to reduce the interior diameter of the furnace to 25 cm, down from the 28 cm ID which has become the most common size for clay wall furnaces in previous testing. The expectation for this experiment was that with air rate reduced to only 72 % of 'low minimum', the yield would also be significantly reduced, down closer to the 3+ kg / <18 % produced in the 2009 - 2010 'Vinland' series. The table above compares various critical aspects of this smelt with the last one undertaken, where the end results are quite similar, despite the great differences in air volumes and burn rates.

One significant difference with an electric blower is that it produces a blast of consistent volume and delivered pressure. Use of a twin bag bellows on the other hand, pluses air delivery, varying both volume and pressure over each individual push of an individual bag, plus differences between left and right side delivery. This cycle occurs about once per second, and is also likely to vary between each stroke, over the working shift of one worker, and certainly different for individual replacement workers over the roughly 4 hours (or more!) required for a full smelt. It may be these changes and inconsistencies are the cause for the great differences in results?

2) 'Vinland' clay

The single test patch was located somewhat beyond the very hottest part of the furnace (around the tuyere), but was positioned at the lower part of the furnace, exposed to the heat of the slag bowl. Although only composing about 15 % of this lower wall area, still this represented enough of the furnace surface to suggest that, even as mixed, the materials sourced at Norstead likely would have survived long enough to complete at least a single smelting sequence. The relative fragility of the clay / sand mixture tested do call into question how well an entire furnace built of this material might function however. Another possibility would be structural components of stone, either supporting a clay liner or even this clay used as mortar. It is clear that the heat of use certainly exhausted any bonding ability of the clay mix, with only a thin inner shell, slag covered and largely fused, from those areas at the bottom most area of the furnace, would survive even the shortest time.

Ideally the next experimental test would be with more careful gathering and fuller preparation of the clay material, with enough gathered to allow for construction of a complete bloomery furnace. This would be followed with a complete drying and smelting sequence using a similar analog, or even better the actual primary bog iron ore available at L'Anse aux Meadows.

Part Six - Moving Forward :

Following morning, extraction side. Note breaks along 'brick' seams

A closer examination of the furnace was made the day following the smelt, as the area was cleaned up and tools put away. It was clear that the impact force of the use of the 'thumper' had resulted in considerable damage to the lower portions of the walls. The wire wraps had held the structure together, but the damage was especially found at the lowest level of the original clay 'bricks'.
It is my estimation that this furnace can be repaired, certainly well enough to allow for a second use. It is however too fragile to allow for any attempt at moving the structure from it's central position in the main smelting area. (It should be noted that at this point there are already two other previous furnaces in place and being retained against a long term observation of natural decay over time. These two experiments hoped to extend into 2031, if the observer's health permits!)

Two additional experiments related to simulating lower volume / Norse bellows use come to mind :
a) Building an air control unit that uses a mechanically driven sliding plate in place of the fixed blast gate set downstream of the electric blower. By using contoured plates, the pulsing nature of human powered strokes could be mimicked. Although long considered, the mechanics have proven daunting.
b) Repeat the experiment using a furnace with a significantly larger internal diameter. Ideally this after a general survey of the existing Viking Age smelting furnaces with an eye to approaching an average historic size. This may prove problematic however, and there are no existing artifact bellows, or even any clear indication of what sizes (so volumes) may have been actually used.
By far the simplest approach would be simply to return to using the actual 'smelting' bellows currently on hand. The primary draw back there is the requirement for the labour pool required just to operate the bellows over the 6 - 8 hours estimated for a full smelt. (The average age of the current smelting team is into their mid 50's!)

Notes :

1) see the guide : ‘If you don’t get any IRON…’

2) Further details on the reasoning behind, and preparation of this analog mix can be found : Ore Analog Composition (blog post)

3) For the report on the 'Wind' portion of the October 2021 experiment :
For a look at the 'smelting bellows' in use :
For the graph of air production during the 10/2021 bellows tests :

4) For a discussion of the aspects of air equipment potentially used during the Viking Age, see 'An Iron Smelt in Vinland' :

5) Reference Lee Sauder & Skip Williams (2002), 'A Practical Treatise on the Smelting and Smithing of Bloomery Iron' :

Unless otherwise indicated :
All text and photographs © Darrell Markewitz, the Wareham Forge.