A Review : 'Iron Smelting in Viking-age Iceland' by Short & Oskarson

 Review :
‘Iron Smelting in Viking-age Iceland:
 a study based on experimental archaeology’
William R. Short and Reynir A. Óskarson
https://www.hurstwic.com/research/iron/index.html

Disclaimer : I had been involved, at the initial start of the Hurstwic iron smelting project, then during the lead up to the 2019 project in Iceland described in this document. I was removed from the project with only two weeks advance notice from the departure date. The reason given was ‘differences in approach’. At the time this appeared to be because of my attempts to communicate individually with other team members, specifically contributing archaeologists. It is worth mentioning that I had previous working relationships with several other team members (iron maker Jeff Pringle / blacksmith Jim Austin / archaeologist Margrét Hrönn Hallmundsdóttir) , and had been instructed to stop any attempt to contact them and others involved privately. Bill Short of Hurstwic demanded total control of this project, needless to state that he and I came to not agree on methods, most especially the free flow of information.  

This critique is presented largely in the same order as the descriptions in the text. All quotes from the report.

On Experimental Archaeology :

    This initial section attempts to validate the approaches used by Hurstwic, without detailing any specifics on the sequence, instrumentation, or record keeping that would be employed. Some of this information will be given through the body of the report, but not in a systematic format, with important data missing. The statement “This approach allows us to make deductive leaps and have confidence in the validity of our ideas.” suggests that established methods employed in Experimental Archaeology are not well understood here. (Specific examples of this as this commentary continues.)
    Two hypotheses are presented :
“…it is possible to smelt high quality iron using tools, techniques, and raw materials known to have existed in Viking-age Iceland. “
“… high quality iron can be smelted using locally-sourced materials gathered close to known smelting sites in Viking-age Iceland. “

On Archaeological Sources :

    There is a long discussion about what is considered by the authors about what they describe as “…a problem we call ultimate authority…”, in this case specifically directed at what is presented as “…the widely held belief that iron smelted in Viking-age Iceland was of poor quality, a belief held by scholars and laymen alike and supposedly backed up by literary evidence.”
    Throughout this report descriptions of iron ‘quality’ are often made. Significantly, there is never any attempt to define what these authors consider ‘good quality’ iron. It would appear to be related in some way to iron used for the construction of swords, as discussion of stories within the Sagas of Icelanders is made, specifically to references of swords bending during combat. (I will return to this later)
    It is problematic that numerous references to the Icelandic Sagas are given (1), but there is virtually no actual descriptions of the archaeological finds at Icelandic iron smelting sites, beyond giving the location names. The statement is made “Yet the excavations fail to reveal the process. How were the furnaces constructed?” The recovery of a number of furnaces at Hals (clay lined turf wall construction) and Skogar (stone base with potential turf upper) is well documented (and cited). A good number of other references are cited in the bibliography, but the bulk of these are only available in Icelandic language versions, so are not easily accessible outside Icelanders.  

On Viking Age Iron Smelting :

    Presented as a very brief overview of the general bloomery iron smelting process, what is presented is generally accurate.
    It is significant however that there is little to indicate what is specifically Viking Age / Norse about the rough description presented. No illustrations of recovered furnaces from archaeology (of which there are a significant number of examples over Northern Europe). This is especially frustrating related to the four Icelandic iron production sites mentioned in the section above,
    The following production values are given : “In order to create a one kilogram finished iron bar, about 2 kg of bloom was needed, created using 4-8 kg of ore and 8-16 kg of charcoal.” There is no indication of where these (questionable) figures are derived. (2)

On Research

    It is noteworthy that the values given for the iron ores are properly corrected from the impossible numbers published earlier by Hurstwic. (3) What is lacking is any discussion of the high variability of primary bog iron ores, certainly between physical locations, but also over time. These factors very important when considering the changes in local environments within Iceland over the last 1000 years.

    Pure bentonite has a melting point of approximately 1200 °C (a figure that should have been mentioned). Available clays can vary widely in melting points, especially as compared to the expected internal temperatures within a smelting furnace in the range of 1350 + °C. This remains a universal problem when building furnaces using locally sourced materials.

“The details of the furnace construction are not clear from available evidence, but we speculated the furnace was a pile of turf blocks with a circular open central shaft lined with clay forming the stack. “ 

    Although Smith’s excavations at Hals are mentioned, that evidence clearly indicates construction using a conical stack with a central shaft. Although there is certainly no reason not to use the flat stack method proposed, it is disingenuous to suggest there are no historic prototypes available. (4)

    The section on ‘Measured Furnace Temperatures’ lacks enough details to consider it hard science. How were the temperatures determined? Exactly where were those sensors placed both vertically and in relation to the air input? It is stated “Additionally, turf provides excellent thermal insulation, as shown by temperature measurements of experimental furnaces suggesting that a turf furnace might operate more efficiently than, for example, a free-standing clay furnace.” Potential high temperature production from burning charcoal is never a concern, outside of the dynamics of available air volumes. Heat * retention * is rarely a problem in bloomery furnaces. Given the continuing discussion in this report of the problem of the low melting point of the local bentonite clay, if anything a highly insulating wall is more likely to promote clay lining failure.  The actual situation is that excess heat radiates off the exterior of free standing clay furnaces, and so thicker walls are more likely to significantly melt / erode. It is also stated “…unlike free-standing clay furnaces which suffer from cracking during repeated heating/cooling cycles…” which is definitely not the case. This may have been intended to mean ‘cracking during the initial drying process’, which certainly can be a problem with poor building technique.
    There is a discussion of modifying the available bentonite clay with the addition of silica, the method of using (basalt based) sand discarded as it is felt that natural sand is not widely available in Iceland. Instead, the source for a silica component is given as ash from burning horse manure : “The horse manure ash samples were high in silica and appeared to show promise as a possible refractory material for a smelting furnace.” Worth noting here is the comment made at the start of the report “We wondered if researchers had fallen into the trap we call modern mindset, an umbrella term denoting instances where our modern-day thinking, ideas, and prejudices interfere with our investigation of ancient times. “ Without any reasonable doubt, the concept of adding sand to clay for furnace construction might have been imported into Iceland, along with the whole mechanism  of iron smelting itself. That silica as a chemical even existed was centuries beyond Norse knowledge, or that burning manure and adding that ash would be a useful component could be anything but exactly an example of the modern mindset trap. 

On Tests and Experiments :

“The tuyere, the pipe which admits the air blast into the furnace through the side, was tested using various positions and various materials for which there is evidence of use in ancient times, including ceramic, copper, and iron.”

    This statement is somewhat misleading, in so much that this section appears to imply that Hurstwic operated in a vacuum, testing unknown elements effecting furnace construction and operation, without any suggestions from others. What is missing here is that the Hurstwic team was provided with training in all these elements long previously researched, tested and documented by others, most significantly the work of the DARC team and myself directly. Bill Short and others from Hurstwic undertook a three day iron smelting in the Viking Age training workshop here at the Wareham Forge in June 2018. This included extensive descriptions of the work DARC had already undertaken on the Icelandic / Hals re-creations (at that date, a total of 8 experimental smelts, work published back to 2007, and under discussion back as early as 2002) 

On Protocols :

    Missing important measurement : interior diameter of furnace at tuyere level. As suggested by Sauder, there is a relationship between ideal air volume delivery and cross section area at tuyere. It should be remembered that the important value for a working smelt is burn rate however (see below)
    Although use of some type of tuyere is mentioned, there is no description what so ever given, other than placement to the furnace wall (base depth and angle). The material it is composed of is not stated, no measurements provided, and it is not seen in any of the images. This is a critical absence, as there is considerable discussion via the known archaeology of what air insert system may actually have been used in Viking Age Iceland.
    It is stated both clay and sand were transported to the working site. No mention of distances involved, hiding behind knowledge of Icelandic locations. It remains an important consideration that far more clay is required than sand in furnace wall construction, most typically only 25 - 50 % of the total. If during the Viking Age, clay would be transported as needed, why not sand?
    Some details of the testing results of various clay to ash silica mixes would be helpful, what determined the 40 ash to 60 clay ratio  (so only 16.4 % silica)? (Again noting that if composed of basalt sand, the amount of that material required would not be punishing to transport.)
It is also worth noting that although the original premise of this experiment was the use of ‘local’ materials, the description of  sources of individual components was in fact scattered over much of western and northern Iceland. So the materials were only ‘local’ as far as being all Icelandic.
    Burn rate is given as 5 minutes per kg (m/kg). Ore charging is stated as initially at 10 m/kg, later increased to 5 m/kg. All of these are clearly averages, over many hours and three different tests. This reduces the value of those numbers. No information is ever given of total amounts of ore that was used in each of the three smelts. This is highly significant, as larger ore volume smelts tend to produce higher yield numbers. 

    The following values are presented :
    •    Smelt 1 = 1.4 kg / 4.7 % yield (ore at 64 % Fe / blower)
    •    Smelt 2 = 3.1 kg / 10.3 % yield (ore at 64 % Fe / blower)
    •    Smelt 3 = .6 kg / 2 % yield (ore at 58 % Fe / bellows)
    These are extremely low (to the point of embarrassment) returns, even more so against the high elemental content of the starting ores. As has been mentioned, if extremely small total ore amounts were added, such low production values might be seen, but that begs the (unanswered) question of just why minimal ore amounts were used. There is striking difference between smelt 1 and 2, for which no attempt at explanation is given.
    There is an attempt to blame the extremely bad results from smelt 3 on the presence of sulphur in the (different) ore used. This is not the effect of sulphur’s presence in an iron smelt, which is primarily an effect of the forging qualities of the resulting metallic iron (brittleness).
    There is no specific information provided about the air volumes / burn rates established for smelt 3. The human powered bellows used is not described in any detail at all, no physical measurements are given. There is no information on pumping rates during operation (certainly to be variable over the number of operators required during the many hours of a smelt). Most likely the poor results from smelt 3 are due to the use of this equipment.

On Results

“ Analysis of the iron from furnaces #1 and #2 showed that it is excellent iron, nearly 100% pure, with an excellent crystalline structure and few inclusions of slag or other impurities.”

    Once again the description ‘excellent iron’ is made, without any definition of what this means. There is a marked difference between the image of one of the bloom pieces shown and the microscopic analysis presented. Clearly the bloom overall is fragmented, contains considerable visible slag and voids. The piece illustrated has been barely compacted after extraction. Contained carbon is described as an ‘impurity’ (more on this below).

Bloom section and compacted bar by Hurstwic from the Icelandic smelt.

 

“We subsequently used period techniques to form a part of one of the blooms into an iron billet, further proving the excellent qualities of the iron.”

    What exactly where the ‘period techniques’ used for compaction here?
    As noted, the reported bloom to bar return of only 50% does not indicate ‘excellent qualities’ to the created iron. Although an extremely small image, the resulting working bar shows surface cracks and possibly slag inclusions, both quite undesirable qualities to blacksmith.

“It was more than good enough for making tools, weapons, or other useful products.”

    Sorry. The initial contention was that the general impression of historic Icelandic iron being ‘of poor quality’ has been specifically joined to iron of suitable characteristics for particularly sword making. Low carbon iron was most likely the preferred product of Norse smelting efforts, the resulting metal is easiest to forge and for processes like hammer welding. The addition of small amounts of carbon (so potentially ‘bloomery steel’) changes the characteristics of the metal. Weapons on the other hand, require the addition of small amounts of carbon within the alloy (0.2 - 1.0 %) for rigidity and hardness. A soft, carbon free blade would in fact be likely to bend in combat use. It also would have little ability to allow or retain a sharp cutting edge.

“It is quite possible that Hurstwic's iron was the first iron smelted in Iceland for many centuries.“

    Incorrect.
    Earlier work by archaeologist Margrét Hrönn Hallmundsdóttir, who not only is cited as a source, but was actually one of this project’s team members, included a number of experimental iron smelts, conducted in Iceland, using stone chamber and turf wall construction, and date back to 2012. 

On Conclusions :

    It is clear that this team did in fact undertake a limited set of bloomery iron smelts in Iceland, utilizing primarily materials sourced from natural sources within Iceland, resulting in some iron being produced. At least one section had been compacted to a working bar. 

“We were unable to falsify our two hypotheses, and so they still stand: evidence suggests that it is possible to make high quality iron with materials and methods known to have been available to Viking-age Icelanders; and that it is possible to make high-quality iron with locally sourced Icelandic materials.”

    Throughout this section, there is repeated use of ‘high quality iron’, yet there is nothing indicated by this report (beyond simple boasting) that this was in fact the case here. The extremely poor yields from these tests do not suggest any kind of reasonable return against the considerable effort involved. So much so that a solid case could be made that if historic iron makers would even consider the methods illustrated, it would be hard not to consider both their skill and product be ‘poor’.
    If a solid definition of what constitutes ‘high quality’ or ‘excellent’ had been given in concrete terms (rather than vague subjective impressions), perhaps the conclusion stated above might have some value.
Any attempt to link these results to the handling characteristics of Viking Age Icelandic swords is questionable at best.

Image taken from the report as published on the Hurstwic web site (viewed directly from that source) https://www.hurstwic.com/research/iron/pix/fig39_bloom1_and_bar.jpg

1) The dominance of the Icelandic Sagas as accurate accounting of historical events, even down to small practical details, in the past investigations by Hurstwic, is clearly obvious throughout their published reports, lectures and videos.
2) Although it could be fair to say that the authors are covering themselves with the additional comment : “The ratios are highly variable and depend on many factors.” The conversion of bloom to bar given at 50% loss is extremely poor work (based on considerable documented work by myself and others), unless the starting blooms themselves are of low quality (meaning fragmented, and / or containing many voids and excessive slag - often seen in small blooms). In comparison, the ratio given for ore into bloom is quite high, up to 50% return given (a figure more typical of very large smelts, or with use of extremely pure iron content ores). Where this number is derived is questionable, especially since the best yield from the actual experiments was reported at no more than 10 %.
3) Originally the same pie chart was presented, only given as pure elemental content - not as the various oxides, the ore was stated as being ’91 % iron’. Not as Fe2O3 oxide only containing in total 70 % iron, so actually at best only 64 % elemental iron. This does still remain an excellent quality ore. Published (as promotion) by Bill Short on to the ‘Iron Smelters of the World’ Facebook group. see commentary : https://warehamforgeblog.blogspot.com/2021/01/truth-in-reporting-sample-iron-content.html
4) see : Markewitz, 2007  : http://www.warehamforge.ca/ironsmelting/HALS/index.html

Note to readers : I present my clear bias right up front.
It is painfully clear to anyone that the Hurstwic report contains more omissions than hard data. Realistically, with so few measurements given, this report documents an Experience - not an Experiment.
I have served as a reviewer for the EXARC Journal for several years now, specifically for submitted articles related to ‘ancient technologies’ I have to state, despite my obvious bias towards this project, that I would not recommend the Journal publish this article in its current form, but instead have it returned for considerable re-writing.

I am personally offended by the refusal of the authors to reference the published work I know full well they based at least their initial understanding of not only bloomery iron smelting in general, but the methods used during the Viking Age, and with Icelandic turf walled construction specifically. Although it is clear that they attempted to weasel out on this lack of credit through the inclusion of a final “The authors regret any omissions in this list.”
When you deliberately leave off any references to training provided, consultation given freely, published documents on which your own work is founded - what do we call that?