September 11, 2023

On data

We had a really lively discussion in our session at the EAA – thanks to all presenters and members of the crowd! – and one comment got me thinking about data and what we mean when we discuss it. I won’t name and shame, but a line of questions from an audience member made it clear that not all geneticists know what archaeologists mean when we talk about our data. I’ve been thinking about that comment for the last bit and thinking about how to respond better than I did on the day.

So, to start, my understanding is (please do correct me if I’m wrong by the way and I’ll write a mea culpa!) that when the majority of lab scientists think about data – specifically the data they analyse – they have some very specific things in mind. Data are in a specific form and format, blots or tables of numbers to be identified and statistically analysed, for example. The materials from which these data are analysed are predictable and known – reagents, cells, samples of specific sorts prepared according to specific protocols.

For archaeologists (and again, apologies for universalising – I’m drawing on my own background and my experience on three continents but I know not everyone does things exactly the same way) this is a little different – data is a fuzzier word. Data in archaeology tends to mean two distinct classes of thing. On the one hand, the material we uncover or recover (from sites, museum collections, etc.) are data. So an excavation profile (the stratigraphy) is data, the soil sample is data, the artefacts are data, the box of flint daggers is data, the landscape topography is data. On the other, the results of our analyses of these things are also data – so my spreadsheet of metrics and qualitative observations of the flint daggers is data, the GIS model of least-cost paths generated from the landscape topography is data, the geochemical results of the soil sample’s analysis are data, etc etc etc. In other words, we tend to consider both our materials and the output of our analysis data. This results, I would guess, from the recognition that both are the product of transformative research – the soil sample becomes data through the process of identifying a site or feature and enacting a sampling strategy on it just as its analysis in the lab reveals geochemical information ripe for interpretation.

The implications of this for us is that there is some flexibility around what can be data – new classes of data emerge when we have new tools or methods of study to hand. As an example, as new dating methods are invented, we start collecting new or different materials (or we collect them differently) because they have become data to us. Radiocarbon dating, for example, prompted archaeologists to start collecting charcoal in a systematic way.¹ There is a sort of recursive relationship between methods and data in archaeology where each helps invent or envalue the other. Another way to put this (in line with social studies of science, but sometimes disconcerting to some empiricists), is that we are constantly defining new things as data – we quite literally make data by delimited specific classes of our materials and naming them significant for analysis. Different traditions of archaeology—from different regions, focussing on different periods, adapted to different site environments—class different things as data, collect material different, and create different sorts of records based on the methods applied in the field and also after.²

One of the reasons it’s fun to be an archaeologist is that the feel is a little unstable – we’re constantly grappling with unruly and complex materials which, to some extent, defy our attempts to categorise them. Our openness to new methods, collaborations, and types of analysis stems from this instability as we try and cope with the complexity of an incomplete record of the past. This is why archaeology is not exactly a science, not exactly a social science, and not exactly a humanity subject, but something messy among all of these.