Wednesday, March 28, 2012

Scatter matters: Bayesian statistical modelling and evidence for overlap between late Mesolithic and early Neolithic material culture in England by Seren Griffiths (University of Cardiff): Review

[** If you like this post, please make a donation to the IR&DD project using the button at the end. If you think the review is useful, please re-share via Facebook, Google+, Twitter etc. **]

Preface:
I am very happy to introduce our third guest writer, Rena Maguire, to the blog. Rena is an undergraduate student at QUB, in her second year. She is currently working on her undergraduate thesis: Iron Age horse harness Y pieces: function, manufacture and typologies.
Robert M Chapple

Being an archaeology and paleoecology undergraduate in QUB Belfast has perks, one of the best being the PCC Lunchtime Seminars organised by our department. I’ve personally found them invaluable to gain insights into sometimes quite obscure areas of archaeology. The guest lecturers are, as one would expect, at the cutting edge of their respective fields.

Not least of these is Dr Seren Griffiths of the University of Cardiff, who gave a talk (March 20 2012) on Scatter Matters: Bayesian statistical modelling and evidence for overlap between the late Mesolithic and early Neolithic material cultures in England. Griffiths has been working with Professor Alasdair Whittle and Dr. Alex Bayliss on an ambitious project, examining links between culture, sites and practices at late Mesolithic and early Neolithic sites throughout Britain. Dr Griffith’s PhD was on the application of Bayesian modelling to dating early sites.

There’s no doubt that the Bayesian modelling has revolutionised dating methods in recent years, as demonstrated by the seminal publication ‘Gathering Time’ (Bayliss, Healy & Whittle 2011) (review: here). Radiocarbon dates, even when calibrated, can be imprecise by a range of hundreds of years. Bayesian modelling checks each year of a radiocarbon date range against other data such as stratigraphic relationships of finds, and in doing so can reduce a date range to a couple of decades instead of centuries (Buck, Cavanagh & Litton 1996).

That’s the science part out of the road! Dr Griffiths emphasised that the main purpose of any refinement of dating methods is to actually know what you want to date. Sounds obvious, but radiocarbon dating is just a number unless you know what to do with the information obtained, and that involves full analysis of taphonomy and association from any given site. Everything is residual, Dr Griffith emphasised, unless you can prove otherwise. This must take into consideration ‘uninformative prior beliefs’ as well as ‘informative’ prior beliefs, to create the Bayesian probability model. Using only what one thinks should be there is smearing the data, and enforcing one’s own agenda on what should be neutral mathematical figures.

To move beyond simply obtaining dates and look at the dynamics of a site, Griffith states, one must produce assemblages of dates on material from related events within that period, and one can only achieve this through associations with the material culture of that society (Bronk Ramsey 2000).

When you’re dealing with the material culture of the Mesolithic and Neolithic, you’re going to be looking at microlith technology. Dr Griffith had specific interest in the microliths of Yorkshire and Humberside, specifically some key sites such as March Hill and South Haw in the Pennines. While the Neolithic package arrived early in the Thames Valley region of Britain, it was adopted much later in the peaty highlands of the Pennines. Dr Griffith pointed out that because of the overlap of cultures, there really is no such thing as a Mesolithic or Neolithic date, or exact point of transition. Both cultures exist simultaneously in close proximity to each other.

March Hill, in Yorkshire, which was excavated in the 1990s, shows continuity of occupation, with Trench 8 specifically containing two hearths from two different periods of time, with rod microliths recovered from each context. However, there is a change from chert rods to flint - the demonstration of material cultural change (Spikins 2003). The similar association with material culture can be shown by the assemblage of microliths found with the Lydstep Pig.

These upland peaty landscapes demonstrate the relationship between landscape features and material culture.  It is within the Pennines that Mesolithic lifestyles persisted when the rest of Britain had embraced the Neolithic package of domesticated animals, cultivation of grains, and pottery production. Dr Griffith presented a view of an eventful Mesolithic history from around 4100 BC, rather than a people clinging to the past.  Close to March Hill Top and March Hill Carr, as well as the South Haw sites people had already adopted the Neolithic way of life. These people were closer to the coast, perhaps more open to outside influences than those in the rugged uplands. At any rate, Bayesian modelling shows the period of 4100-4000 BC as a period of dynamic transitions in Britain, and demonstrates the speed with which cultural changes occurred after this overlap period.

I won’t pretend that I know very much about Bayesian modelling. Therefore I hope I have done some form of justice to the mathematical side of Dr Griffith’s presentation. What I was impressed by was the acknowledgement of our pre-conditioned ideas about the transition of the Mesolithic to Neolithic period. Call it iconoclasm, but I rather relished the idea of no preconceptions of dates and relying purely upon material evidence, and how it relates to dates gathered. I for one came away from this seminar resolved to purchase some books on Bayesian theory, which  now made an awful lot more sense than it had before!

References:

Buck, C. E., Cavanagh, W. & Litton, C. 1996 Bayesian Approach to Interpreting Archaeological Data (Statistics in Practice). London.

Spikins, P. A. 2003 Nomadic People of the Pennines: Reconstructing the Lifestyles of Mesolithic People on Marsden Moor. London.


Note:
There are a number of on-line introductions to Bayesian modelling for the archaeological community. These include one produced by Dr. Rick Schulting for the INSTAR workshop at QUB [here] and another one, produced by Thomas S. Dye [here]


[** If you like this post, please consider making a small donation. Each donation helps keep the Irish Radiocarbon & Dendrochronological Dates project going! **]

If you’re going to do any shopping through Amazon, please go via the portal below. It costs you nothing, but it will generate a little bit of advertising revenue for this site!