Making Sense of Analytical Data

After the inital excitement of the arrival of new material in the lab, curiosity had to be curbed and the main task of the day tackled. This task was to process and interpret anayltical data acquired last week during many days work on the SEM. I use many different analytical techniques to investigate the more important archaeometallurgical residues passing through the lab – and the analytical SEM is one of the most useful.

BSEM Roman smelting slag

Backscattered electron image of a tapped Roman iron smelting slag. The field of view is 2.5mm.


The backscattered electron images reveal compositional contrasts through their grey scale. In this image the dominant phase, appearing pale grey, is fayalite (an olivine mineral, approximately Fe2SiO4).

Across the centre of the image is a discontinuity, produced by the chilling of the surface of an individual lobe of slag as it flowed from the surface and cooled in the air.

The crystals are large, suggesting the slag cooled slowly, and the lobe margin is not marked by the development of much iron oxide, so this example probably cooled right in the mouth of the furnace.

As well as producing these images, the analytical SEM also permits chemical microanalyses from tiny spots or areas of the sample.

The second backscattered electron image shows a tiny detail of the first image, with the location of microanalyses.

Detail of Roman iron smelting slag

Detail of Roman tapped iron-smelting slag. Field of view is approximately 0.17mm.


The instrument provides the chemical analyses, but they then have to be recast as mineral formulae – and that was today’s task. With many hundreds to do that was a substantial task in front of the spreadsheet. Gradually a picture emerges of the overall composition of the slag and of its constituent minerals.In this instance, the slag proved to be typical of residues produced during the smelting of iron ores from the Forest of Dean. That is a useful result in itself, allowing one aspect of the economy of this Roman settlement to be understood. As other samples from the same site are interpreted further details will emerge – permitting reconstruction of the yield and efficiency of the furnace as well as aspects of the technology itself.

Spreadsheet of chemical data

Processing microanalytical data, to convert the microanalyses into mineral formulae.


Archaeometallurgical residues provide a very direct link back to a particular occasion in the past, when an artisan did a particular job in a particular way. The waste material provides key evidence for that moment in time. Although studying the waste, rather than the product, might seem perverse, there is often a richer set of evidence about hte nature of the process to be gleaned from the residues than from the artefact. Crucially, the residues also typically remain close to the site of the activity, whereas the products were dispersed after production and may not be able to be linked back to their point of origin.

Careful investigation of such archaeometallurgical residues may allow us to come as close as we ever could do to looking over the shoulder of the Roman smith at his work.

Data analysis in the afternoon

Some people say the morning is the best time to write, but I like the afternoon.  In the morning I’m too distracted by all the various to-dos that I know are on my list for the day; I find I’m better off getting some of those things done in the morning and then plopping down in front of the computer after lunch. The only problem with this schedule is that it seems many others use afternoons as their errand time.  So while my phone, email, etc tend to be blissfully silent in the morning, in the afternoon, if I need to leave my phone/email on for some reason (or if I just forget to turn them off) it’s a constant stream of interruptions.  So it has been this afternoon.


My plan for the afternoon had originally been to finish up an article on one part of the Navajo project – it’s almost there.  But then I got several emails/phone calls, all about different important matters that don’t take much time to address but which I did need to deal with.  Unfortunately, I don’t deal with distractions at all well while writing; I really need a block of time in which to concentrate.  So I abandoned ship on finishing the article today.  Maybe over the weekend.


Instead, I turned to data and statistics.  The beauty of this kind of work in this situation is that it’s something I can do with interruptions – in fact, I find interruptions to be useful.  I can keep thinking about a data problem in the back of my head while dealing with something else.


So today when interruptions derailed my writing, I turned to my Spanish project.  Earlier this summer, I was in Valencia, Spain, looking at a zooarchaeological collection of leporids (or rabbits) from the site of Cueva de Nerja.  Now, it’s time to take those data and figure out what they mean.  My question in looking at these rabbit bones has to do with how the rabbits were being hunted.  Did the prehistoric inhabitants of Nerja take these on the landscape?  Or did they hunt them using a mass capture technology, such as netting?  The way to answer this question is by looking at the demography of the rabbits in question – are there lots of young rabbits, or mostly older ones?  More males or females?  Are there changing patterns through time, and if so, are those patterns statistically significant?