After work at the end of 2002 to lower the base of the furnace and the blowhole by 20-30cm, the more difficult task of lowering the top of the tap arch has been started. With only a few days to Science Week, it will be touch-and-go whether the new clay will be dried sufficiently. The tap arch is supported by a sandstone slab suspended partly on steel rods; it remains to be seen whether this will be able to withstand the temperature.

The last jobs before the first smelt of 2003 have been completed. A hand-cranked smith's blower has been renovated by Toby and Aaron and has been mounted to blow alongside the main bellows. Tests suggest the blower on its own can supply up to 1000 l/min. The major alterations to the tap arch appear to have dried sufficiently.

After a year of no smelting, smelt 22 finally took place. The plan for the day was to work with much higher blowing rates than we have done previously, in an attempt to apply the methodology of Sauder & Williams (see Historical Metallurgy, 36, 122-131 and the Rockbridge Bloomery website). However, during testing of the blowing system prior to the start of the smelt it became clear that the fan blower added little, if anything, to the overall blow when attached to the furnace, and the efficiency of the bellows was probably reduced when connected to the water-pipe manifold system. The smelt went ahead as planned. The furnace was run for 1h:35m before charging with ore, with the start of charging being just after ignition of the top gasses. 16.8kg of ore was added in 2.8kg batches (each accompanied by 2.8kg of charcoal) over the following 2h:05m. This was equivalent to a burn rate of 6.7kg/hr; approximately one third of that in Sauder & Williams' furnace. The furnace was left to burn out the last of the fuel overnight.

The day was a busy one, with many school visits.

Alongside the smelting, the day was marked by our first foray into charcoal making in an oildrum. The design of the drum should allow us to make a batch of charcoal within a day - but learning to operate the drum may take time. The burn appeared to start well, but it went out by mid-afternoon, possibly because of a too conservative approach to allowing air to reach the inlets. We will relight the kiln on Saturday.

The slag mass from smelt 22 was removed and examined. As expected, there appears to be little iron present. The "burr" region of interaction with the furnace wall below the blowhole was very well developed and came out in one piece. Some fluid slags had dripped below the mass, but none had reached the furnace floor. Removal of the slags caused some damage to the inner face above the tap arch and below the blowhole, but it was decided to go ahead with smelt 23 without major repairs to these areas. Drastic action to improve the air supply needed to be taken, and a "bouncy-castle" blower has been booked for tomorrow (thanks to Gary & Raphaella Fieldhouse of Body Canvas). This will provide the opportunity to take the experiments to a new level - but connecting the blower to the furnace will be a challenge!

The second of the smelts for National Science Week 2003, smelt 23 saw a radical change in experiment design. This was the first smelt undertaken with a mechanical blower. Experimentation showed that the output of the blower dropped drastically with load, and the blower was harnessed directly to the furnace, without requiring any release of air, contrary to our expectations. The estimated blowing rate is approximately 1000 l/min. The smelt progressed very smoothly, with charging starting at 1:08, and with a 15.2 kg/h burn rate, consuming 95kg of charcoal and 22.4 kg of Sishen Ore (charging 1:08 to 2:10).

Batches of slag were tapped and recycled at 2:45, 3:17 and 3:37. The slag flowed particularly well on the last of these attempts, and perhaps other taps should have been made. After about 4:00 the blowhole was occluded on a regular basis, involving several attempts at clearance. The blowing wall of the furnace became quite hot on the outside at this time. The burn-down phase started at 4:21, but the burn rate dropped markedly and at 6:15 the blast was switched off and the furnace allowed to burn down on its own. By this time the extent of the bad damage to the wall above the blowhole was becoming clear. A major repair will be needed.

The mass in the base of the cooled furnace was removed - not without some difficulty. The mass comprised (from the bottom up) a thin firm slag coating to the bowl of the furnace, which was filled with variably vesicular and charcoal bearing slags, capped with a pool of what had clearly been very fluid slags. Towards the blowhole side the pool was overlain by a large void, in turn capped by the bloom, which lay directly on the slag distally and away from the tap arch. The bloom appears to include about 5kg of iron, but was covered by a large quantity of very pale green-yellow slags, probably associated with the late failure of the blowinghole-side wall.

The upper image shows the mass as viewed from the tap-arch side. The mass is 0.24m tall. The void separating the bloom from the slag-filled bowl is visible, together with the slag "puddle" just below.

The lower image shows a subsequently-cut section through the bloom, showing a good solid structure with very little included slag. The coin for scale is 22mm diameter. The total iron content of the bloom is estimated as between 5 and 6kg.

Preparations were made for smelt 24 - which was to happen away from St Fagans, in a temporary furnace. The furnace was constructed with an internal framework of bricks; firebricks towards the base and in the tap arch, with housebricks higher up. The structure was infilled and lined with a brick clay. The outside was coated in a locally-dug clay.

The furnace was only to have a single night in which to dry before smelting tomorrow. A fire was set in the furnace for a few hours to assist this process, but it was considered undesirable to dry the furnace too quickly today.

The structure seems very strong, and is slightly more regular than the normal furnace, with an almost constant internal diameter of approximately 31cm.

Smelt 24  was undertaken. The parameters for the smelt were to be the same as for smelt 23 - with the exception that it was to happen in a different furnace (i.e. in contact with a different clay, and using a slightly larger bore blowhole). The blower, ore and technique were to be the same. Because of various factors beyond our control the preheat phase took longer than before, with ore charging commencing at 1h:45m and continuing for 1h10m, with a burn rate of 14kg/hr. Attempts to tap the slag at 3h:20m, 4h:40m and 5h:20m all met with failure. Highly viscous flows were seen descending from above the tap arch - but free flowing slags were absent. This led to concern that the furnace lining was too refractory and that not enough silica was available to the reaction. This fear was furthered by the discovery of a bloom very firmly attached to the wall, which took a considerable amount of time to remove. The raw bloom mass weighed 5.2kg, but it seems likely that a large proportion of this will prove to be slag. Nonetheless, iron is clearly present, and an attempt to smith the bloom will be made tomorrow.

The temporary smelting furnace was reconfigured as a smithing hearth by the removal of the upper courses of bricks. Bloomsmithing commenced with three reheats over the first 55 minutes. These reheats resulted in a rapid loss of slag from the impure bloom, which dropped in weight to 3.6kg. The bloom then broke in two in the hearth with the retrievable fragment weighing 1.75kg after reworking 4. The reworking continued with a wooden anvil and mallet for a further 16 reheats over a period of 2h:12m before reworking switched to a steel hammer and anvil (by which point the bloom weighed 1.28kg. Five more reheats were undertaken during the following 1h:5m at which point the bloom disintegrated in the hearth and only 580g of small iron scraps were recovered.

This was a very disappointing end to the weekend's smelting, but only serves to emphasize the significance of having the correct lining composition; using a highly pure ore demands that silica be available from the lining to provide a fayalitic slag. the most likely explanation of our problems this week end seems to be that the clay lining was far too aluminous, and probably also too thinly spread over the highly refractory firebricks. Although the general construction of the furnace was stable and strong, these matters will need to be addressed before attempting a similar style of construction again.

To lift the mood a little, the last hour was spent bloomsmithing the major part of the bloom from smelt 23. The original approximately 7kg mass had one end removed by angle grinder to reveal a section and then a further 20mm of the bloom was cut off to archive (940g). The remainder of the bloom (approx. 4kg) was then reheated and smithed through several workings over a period of approximately 80min. The bloom lost a fragment of 280g late in the process, but the product was a part-worked bloom, reduced to a bar 65 x 55 x 160mm, weighing 2.5kg. Final smithing of this bloom will be attempted on another occasion.

Work on the new bloom reheating hearth was started. The structure is fonded on the clay stump of our original smithing hearth.

The bowl of the hearth was dug into the old clay, with walls to be built of clay, cored with bricks to provide stability (using the "rapid-build" style adopted for the temporary furnace for smelt 24). The clay is a fine clay bought in by the museum for another purpose; its potential for use in hearths/furnaces will be evaluated through its performance in the reheat hearth. It is certainly a nice material with which to work. The blowhole employs a section of offcut from scaffold pipe.

Space inside the smelting shelter is limited, and there will be problems in mounting the bellows for the hearth if the smelting furnace is in use.

We have now taken delivery of an electric blower. The success of smelt 23 and the "qualified success" of smelt 24 have made it clear that having a capacity to blow at high blowing rates with an electric blower is highly desirable for the experimental series, despite the better "visitor experience" in seeing the furnace blown by hand. If real success can be had with the electric blower, then we can plan for trying to provide those blowing rates by hand at some point in the future.

The blower is 0.55Kw unit from Bee-Tee Inflatables, essentially the same unit as the one hired for smelt 23.

Work on the structure of the bloom-reheating furnace is now complete.

Smelt 25 was undertaken under atrocious weather conditions and was filmed for a television programme.

It took 1h 40m from firelighting to the point at which the top gasses (mainly carbon monoxide) ignited. Normally we would start the charging phase at this point, but in this smelt we delayed slightly (for reasons of the filming), and didn't start adding ore for a further 40 mins. The charging phase extended over a period of about 1h 40m, and involved adding 6 batches of ore and charcoal. Following charging, the tapping phase lasted 1h 30m, with three tappings - of which only one produced free-flowing slags. The slags were recycled into the top of the furnace. At the end of the tapping phase we added one last batch of ore. In this smelt the burn-down was very protracted - partly because of the interruptions due to filming when the blast was switched off for a total of about 1h 20m. We abandoned efforts to burn-down after about 3h 30m. The burn rate during the main part of the reaction was approximately 10kg of charcoal per hour. This is much lower than we would have expected; on the the last run of this furnace, the burn rate was approximately 15kg per hour. It is important that we find out why this happened before we smelt again. Very little had been changed since the last smelt - the only major variables were the different blower (we were using a different blower, but of the same model, compared to last time), the weather, the charcoal (had it got damp perhaps?) and the filming. There were delays imposed by the filming - but these did not happen at crucial times and it seems unlikely that this could have produced the low burn rate. Further investigation is needed! I guess that either the blower wasn't performing well, or that the charcoal was damp.

The second part of the days operation was to do some bloom smithing. This was barely started before nightfall. The part-smithed half bloom from smelt 23 was chosen for continued working. The part-forged billet lost a small fragment from one end (approximately 130g), but the rest compacted well. After approximately 10 heats the billet was reduced to a cube weighing 2.09kg (therefore having lost a total of 0.41kg today and being just a little over half its original weight). The smithing hearth performed well, although the smiths felt it was at the limit of its capabilities, in terms of both the size of billet being forged and the maintenance of temperature when hand-blown with our old large smith's bellows.