Our Experimental Archaeologist Therese Kearns explains a recent experiment conducted as part of the Accidental and Experimental Archaeometallurgy conference at the Ancient Technology Centre in Dorset.

Butser colleague Fergus Milton and I recently attended the Accidental and Experimental Archaeometallurgy conference organised by the Historical Metallurgy Society in the Ancient Technology Centre in Cranbourne, Dorset.

The conference programme was divided into oral presentations and practical experiments covering ferrous and non-ferrous metallurgy.  This meant that there were furnaces of all shapes and sizes dotted around the site and a buzz of enthusiasm for high-temperature technologies, which makes these gatherings so wonderful. 

The last time I attended this conference was in 2010 when I was involved with iron smelting experiments, but this time we decided to take a leap into the unknown and experiment with the production of brass.

While we were familiar with the theory of brass production by the cementation process, we had never tried it practically, so we were excited by the steep learning curve that awaited us!

Brass is an alloy of copper and zinc and has a yellow colour much like gold.  It was prized in antiquity not just for its colour, but also because it was harder, stronger and more malleable and ductile than bronze (an alloy of copper and tin). 

The Romans were producing brass using the cementation process by at least the 1st century BC.  Initially, it was used for coinage, but it soon largely replaced bronze for the manufacture of decorative metalwork.  Traditionally, the introduction of brass to Britain was associated with the arrival of the Romans, though there is some evidence of brass metal from late Iron Age contexts.

The metallurgy of brass is fascinating and more complex than other alloys such as bronze where the alloy constituents (copper and tin) are mixed together in their molten state.  Zinc melts at 419°C and vaporises when heated above 907°C which is below the melting point of copper (1083°C).  This means that zinc and copper cannot be in a molten state at the same time so they can’t be melted together.  In antiquity, the brass manufacturing process involved first producing zinc oxide by crushing zinc carbonate ore (such as calamine) and layering it with powdered charcoal (which acts as the reducing agent) then roasting it at around 700°C.  The carbon combines with oxygen which is released as carbon dioxide gas (CO2), leaving behind zinc oxide (ZnO).

The zinc oxide was then placed in a crucible with charcoal and granulated or finely divided copper.  The zinc oxide reacted with the carbon inside the crucible to produce zinc vapour which diffused through the copper while in its solid state.  It was crucial that the crucible was sealed, otherwise, the zinc vapour would be lost to the atmosphere.  It was also important to control the temperature so that it remained below the melting point of copper to ensure the copper remained solid until the maximum diffusion of the zinc vapour had occurred. If the copper melted, it would sink to the bottom of the crucible and reduce the surface area available to absorb enough vapour to form brass.  

Our experiments were inspired by metalworking debris excavated at the Westward House area of Fishbourne Palace in the 1990s.  Among the excavated finds were fragments of crucibles and moulds which were used for making and casting brass.   

The fragments were from two distinct crucible forms – handmade triangular shallow bowls – a type well known from late Iron Age contexts, and small wheel-thrown jars with an extra outer layer of ceramic material, examples of which were also found in Roman contexts in Cirencester, York and Silchester.

As these experiments were our first venture into practical brass production our main aim was to  get a feel of how the process might have worked, explore different crucible types and understand the working temperatures required for the process.

We had hoped to use a thermocouple to take regular temperature readings, but unfortunately, the thermocouple malfunctioned, so we had to do without apart from one run when we were able to borrow from a colleague.

All the experiments were conducted in a shallow bowl furnace with air input via a tuyere extending to the centre of the furnace pumped through a small set of goat-skin bellows. 

We had prepared a range of crucibles of different shapes and sizes, all hand-built, and some similar to the triangular forms seen in the archaeological assemblage from Westward House.  The crucibles had to be strong enough to hold the metal and resist the heat but also needed to be sealed to avoid the loss of zinc vapour.  We were unable to procure any zinc ore but were happy that zinc metal would suffice for these initial experiments and would not detract from our broad understanding of the process. 

The learning curve was every bit as steep as we had expected and although we were experienced in smelting copper and casting bronze in a similar furnace where temperatures above 1000°C are required, the requirements for this process were very different.  Here, the challenge was to achieve temperatures above 907°C but under 1083°C and maintain it for as long as the process required.  How long was that?  We had no idea!

For our first attempt, we used a crucible with an indented lid (figure 2) which we hoped would be self-sealing.  The crucible was charged with 50g of copper prills and 15g of granulated zinc.  When we thought we had reached around 1000°degrees (which we gauged by the colour in the furnace) we continued to bellow for 12 minutes at which point we stopped and opened the crucible to have a look inside.  The zinc was still solid suggesting that we had not reached a high enough temperature and certainly not high enough to penetrate through the crucible fabric.  We continued to bellow trying to ensure that the heat was being distributed right across the furnace and into the crucible wall.  After a further 25 minutes, when we opened the crucible, it was clear from the colour of the metal that some brass has been produced, however, we had also melted the contents of the crucible and a solid ingot weighting 63g lay at the bottom.  The exciting thing was that the ingot was clearly brass on both sides, though we have yet to determine how deep that penetration goes.

In total, we ran four experiments using crucibles of different forms and fabrics and different methods of sealing lids, using clay, dung, leather and sheep's wool.  We had varying success on each attempt and in hindsight, I believe that we were being too cautious about the temperature control after we melted the entire contents of the crucible on our first attempt.  In subsequent attempts, it was clear that we had created some brass, but we opened the crucible too soon, just as the vapour was just starting to diffuse into the copper but before the process was complete. 

On the one occasion when we did have access to a thermocouple it showed that we reached 1160° C in the furnace after bellowing for six minutes at a rapid rate, when bellowing stopped the temperature decreased by consistent jumps of 100°C, which showed us that to maintain a temperature just over 900°C we needed to slow down the bellows dramatically. 

We learned lots of lessons from these experiments and will be much more informed for our next attempt – so watch this space!

Figure 2 A lidded crucible ready to go into the furnace, and the resultant ingot with the distinctive brassy colour visible on the underside.