Sunday, June 30, 2013

Bronze age bellows a la Brenda Craddock

Brenda Craddock taught an excellent unit on making primitive bellows on the second day of the Butser archaeometallurgy course, and this is an attempt to capture some of that excellent teaching.

The bellows pattern used was referred to locally as a "Roman Kite Bellows", although I've been unable to find that term used elsewhere. Brenda confirmed that "No one actually knows what bronze age bellows looked like.", but her design is intentionally "As simple as possible" and based on "later descriptions and depictions" of them.

The bill of materials is relatively short:
  1. one large tanned goat skin
  2. one hollow nozzle (we used a segment of bamboo)
  3. two short, smooth, regular, handles (dowels work)
  4. leather awl with thread
  5. a few feet of rough twine
  6. (optional) gelatin glue
The bellows pattern (figure 1 below) itself was designed to fit into an animal skin with maximum utility, with the narrow bit to the left terminating at the scruff of the neck, and the rounded section on the right straddling the tail and haunch. The rest is made as broad as the hide will allow. Caveat: I was unable to trace her pattern directly, so this is a projection from some images captured during the class.


General notes

  • The bellows are produced such that the handles have the rough/suede side of the hide on their outside, so that they seal suede-against-suede rather than smooth-to-smooth. This gives a better seal, which is essential.
  • The bellows can be fashioned such that the seams are on either side, although in this class all the bellows we produced had the seams on the outside.

The foot

See figure 2a. The hide is bent slightly along the A-B line, then points C and E are brought together. The edges A-C and A-E are sewn together starting from point C/E and progressing toward A, so as to avoid any misalignment. Then points D and E are brought together, and sewn as above, from the point D/E toward B. Lastly, a small triangular gusset is sewn into the V formed by the edges CF and DG. The result should look something like figure 2b.

Fig 2a Fig 2b


The snout

See figure 3a. The hide is curled around the A-B line, joining point C with E and D (roughly) with F. Sewing proceeds from point C/E toward point D/F, so that the joint at the handle tabs is clean and any discrepancies occur at the snout. Figure 3b shows the curl and the point from which to sew.

Fig 3a Fig 3b


The handles

As we mentioned before, the hide is wound around the handles such that the rough side is on the outside to provide the best possible seal when used. Once the handle sleeves have been sewn, the handles are inserted most of the way into them, and the bottom of both sleeves are pinched together and sewn flat to prevent air from escaping from the joint. Figure 3b above shows the measuring for this process. Figure 4 shows the bellows at this stage.

Fig 4


The vent

The hollow vent for the bellows is placed in the "snout" of the hide we have sewn. It may be liberally coated with gelatin glue, as may the inside of the snout. Then a meter of twine is wound about the joint between the two. The twine is laid so that its midpoint is beneath the snout/vent joint (see Fig 5a), then wound from both sides. It is single-knotted at least every other winding to prevent slippage, then tied securely when the ends are reached. Fig 5b shows the completed tie. Note that there is considerable linear coverage along the joint.

Fig 5a Fig 5b



Here is Ryan Watts, a classmate, using the constructed bellows:


We measured the approximate air displacement of a single push of the bellows and arrived at a volume of 1.4 liters. With an average of 1.25 pushes per second over an hour long session, we get an airflow of 105L/min. Vigorous pumping increases the pushes per minute but usually decreases the volume per push somewhat. It would be extremely difficult for someone to sustain over 200L/min for an hour.


  • The airflow is easily sufficient for artisinal production of gold, silver, lead, mercury, copper, and tin. It is doubtful whether a single bellows of this type would be sufficient to smelt iron.
  • Given a good leather awl and all the materials required, a bellows of this type could probably be constructed by an experienced craftsman in less than two hours.
  • The materials required could largely be taken from a single animal (Brain-tanned hide, gut/sinew thread and twine, bone nozzle)
  • Considering the number of materials, parts, and processes required to make them, a bellows might be among the most complicated tools on a bronze age farmstead.

Friday, March 29, 2013

Upcoming events

May 24-26: Ontario, Canada
Darrell Markewitz, author of Hammered Out Bits, is teaching an iron smelting class, where I hope to smelt my first iron ore.

May 31-Jun 3: Hampshire, UK
Fergus Milton at Butser Ancient Farm is teaching a four day archaeometallurgy class covering copper and bronze smelting. I hope to walk away with a cast axe head.

Jun 14-16: London, UK
The Historical Metallurgy Society is having their 50th anniversary conference.  I have just paid my membership and registration fee so that I may attend.

I also plan to visit Malta and Morocco, so if anyone knows of any events related to early mining or metallurgy in or around those places, please let me know!

Wednesday, January 9, 2013

Success, sort of.

A second three hour high-heat smelting reduced the contents of both crucibles completely. The copper came out as a single large button. The tin was different. There was one small button (5g) and a three layer "cake" of slag... a light grey very granular layer was the bulk of it. Atop that was a dark grey layer and atop that were little balls of tin, like a froth, embedded in the borax.

In researching the tin, I find that tin forms a significant amount of tin silicate slag, given the chance. Historically quicklime was added to displace the tin from the silicon. I will try that next time. 40% of the mass of the cassiterite in quicklime is the proportion recommended.

Saturday, January 5, 2013

2013 begins: Smelting in a kiln

I got a Cress kiln a while back. I refitted it with a (much, much) longer cord, and here's what it looks like now:

I tried smelting copper and tin in it today, using crushed charcoal as a reducing agent and borax as a flux and sealant.

Crucible 1:
100g cassiterite
 10g powdered charcoal
  5g borax
Crucible 2:
 70g malachite
 10g powdered charcoal
  5g borax
Both crucibles were covered by ceramic discs that fit loosely.

I heated the kiln on "low" for about 30 minutes, "medium" for about 30 minutes, and "high" (dark orange heat) for 2 hours, then reversed the process for cooldown, for a total of about a 4 hour total firing. When I pulled them out, imagine my surprise finding only partially decomposed ore, and plenty of charcoal left. The copper was considerably further along than the tin in terms of decomposition, but neither had a substantial lode of purified metal.

Tomorrow, I shall extend the time for full heat from 2h to 3h and see what's what.