|
Archaeometallurigcal research is taking place on several projects.
The most significant is the investigation of early smelting technology in the Bristol Channel Orefield. This research entails bringing together site-based research on the chemistry and mineralogy of slags, fieldwork evaluating smelting activities (particularly through geophysics) and the development of numerical techniques to investigate furnace reactions. The key issues for the understanding of early smelting in the region are associated with how the slag-tapping bloomery furnaces operate; given that much of the slag is ultimately derived from the furnace lining (the ores are mostly of very high grade), how does the reaction of charge and furnace take place? Can we recognise variations in the process, either in space or time? How does control of the slagging process influence metal compostion and yield?
The biggest component of the investigation of iron-making in the Bristol Channel Orefield is The Miskin Project, which, as documented on other pages, aims to track changing utilisation of a single set of natural resources through time.
The programme of experimental smelting is also designed to feed directly into the understanding of the process of smelting Bristol Channel Orefield iron oxide ores, but drawing from practical experience.
Another, separate, area of interest is a class of slags which commonly appears amongst archaeological collections, but which is very hard to interpret. These slags are often low-iron, and may just be vitrified clay, but which are sometimes rather phosphatic. The site at Bornish, South Uist, yields large amounts of slags, most of which may not be metallurgical, but which arise from reactions of hearth lining and fuel ash in lower temperature hearths under certain circumstances.
The study of early iron-making in the Bristol Channel Orefield through analysis of raw materials and waste-products from early smelting sites is providing new techniques for provenancing iron sources and for understanding smelting technology. Sites being investigated include:
Romano-British tap-slag
Frocester Court Villa, Glos.,
sample FC009.
Backscattered electron image,
field of view 100µm wide.
Bright: wustite
Intermediate: olivine (Fa 86, Fo 5, K 9)
Dull: glass
Much of this work was undertaken by Gary Thomas as a PhD thesis, supported by a Sir Charles Wright Scholarship from Cardiff University. We developed a new graphical technique for investigating bloomery furnace mass-balance which allows us to use all elements analysed in the ore, slag, lining and charcoal, rather than simply solving the minimum number of simultaneous equations. This technique allows us to provide an estimate of iron production (and in later furnaces at least this will be a good estimate of bloom size), slag production and furnace efficiency.
This example shows the graphical solution for selected major elements for materials from the probable late Roman smelting site at Miskin, School Road. The y-axis represents the concentration factor required to elevate elemental concentrations in the model melt to those observed in the slag. The x-axis represents the proportion of ore in the model melt. This graph is for a fuel-ash contribution to the model melt of 1.8%. The best fit solution lies at E = 1.35, ore = 94%, lining = 4.2%. In such a furnace 2 kg of iron metal would be produced by 9.3 kg ore, 0.4 kg furnace wall and 0.18 kg fuel-ash. The process would produce 7.4 kg slag and would have an efficiency (iron in bloom / iron in ore) of 56%.
Investigation of the development of ironmaking around the iron ores of the Vale of Glamorgan has included excavation of a site of Roman age near Miskin, and field/geophysical survey of three nearby late-Medieval/early post-Medieval water-powered sites. These sites form part of the detailed Miskin Project. Although work on the Medieval sites is at an early stage, the mass-balance calculations for furnace operation derived from slags from these sites indicates a very different technology from that seen on the nearby Roman sites. The models indicate a process with a high efficiency, and the size of the slag cakes indicates a process with a high throughput. These data are entirely compatible with the documentary evidence for yield (over 67%) and bloom size (50kg) obtained at a local bloomery in the 1530s.
Magnetic gradiometer survey of earthworks associated with a probably late Medieval water-powered site near Miskin, Glamorgan. Linear anomalies in the centre of the survey correspond to high concentrations of slag within an earthwork, possibly connected to a dam further north. More irregular anomalies to the east represent slags upcast from a 19th century stream-cut (bounding the east side of the survey area), dug through the probable earlier pond.
Geophysical, topographic and trial pit surveys have been conducted over the site of medieval bloomery to the south of Iron Acton, South Gloucestershire. A large, but degraded slag dump has been located, with over 100t of slag remaining. The slag has largely been redistributed downslope of the original dump site. Natural downslope movement was probably exacerbated by post-medieval quarrying of the slag for track surfacing.
Two other smaller bloomery mounds, now almost destroyed by ploughing, have been located to the east of the village.
It is hoped that further investigation of these sites will yield new information on the age and structure of the iron industry in S Gloucestershire.
Contoured plot of the density of slag (in kg/m3) recovered from test-pitting a layer representating degradation of the original slag dump, now without any topographic expression.
A variation on typical slag textures and mineralogy is provided by the study of the Norse period slags from Bornish, S Uist. These unusual slags were produced from peat-fired hearths formed in highly calcareous sand of the machair.
Hearth slag, Bornish, S Uist, sample BOI.1.
Olivine has nucleated on iron oxide (bright) and shows zoning from chrysolite (Fa 22 Fo 78) cores (dull) to hortonolite (Fa 55 Fo 45) (intermediate). The groundmass is a glass.
Backscattered electron image, field of view 100µm wide.
Hearth slag, Bornish, S Uist, sample BOI.2.
Olivine (bright) is of reduced importance and is zoned from chrysolite, Fa23 cores to hyalosiderite Fa48 rims. The dominant ferromagnesian mineral is the clinopyroxene augite. The groundmass is a glass.
Backscattered electron image, field of view 50µm wide.
Close to iron-rich inclusions are small regions of olivine-dominated mineralogy (above), but 300µm away clinopyroxenes are dominant (left). Much of the slag shows little influence of added iron and is dominantly partially melted sand.
Magnetic spheroid, Bornish, S Uist
Scale Bar 1mm
Micro-residues from Bornish include magnetic slag spheres. These are often extremely similar to the well-documented slag spheres produced by the expulsion of slag from fire-welding. The key feature of the Bornish spheroids is that they are demonstrably associated with a corn-drier! Fine-grained residues at Bornish appear to have been generated through the use of iron-rich peat, bearing wind-blown sand grains, as fuel; they are not metallurgical!
These micro-residues provide the clue to the interpretation of the macroscopic slags. They too were generated in non-metallurgical processes. The slags are associated with large hearths dug into the sand of the machair; reaction between the highly calcareous sands and the iron-rich peat generated conditions for production of sinter.